diff --git "a/89FLT4oBgHgl3EQfti_6/content/tmp_files/load_file.txt" "b/89FLT4oBgHgl3EQfti_6/content/tmp_files/load_file.txt" new file mode 100644--- /dev/null +++ "b/89FLT4oBgHgl3EQfti_6/content/tmp_files/load_file.txt" @@ -0,0 +1,1007 @@ +filepath=/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf,len=1006 +page_content='Layout-aware Webpage Quality Assessment Anfeng Cheng∗, Yiding Liu∗, Weibin Li∗, Qian Dong, Shuaiqiang Wang, Zhengjie Huang, Shikun Feng, Zhicong Cheng and Dawei Yin§ Baidu Inc.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', Beijing, China {chenganfeng01,liweibin02,wangshuaiqiang,huangzhengjie,fengshikun01,chengzhicong01}@baidu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='com liuyiding.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='tanh@gmail.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='com,dq22@mails.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='tsinghua.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='edu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='cn,yindawei@acm.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='org ABSTRACT Identifying high-quality webpages is fundamental for real-world search engines, which can fulfil users’ information need with the less cognitive burden.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Early studies of webpage quality assessment usually design hand-crafted features that may only work on par- ticular categories of webpages (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', shopping websites, medical websites).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' They can hardly be applied to real-world search engines that serve trillions of webpages with various types and purposes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In this paper, we propose a novel layout-aware webpage quality assessment model currently deployed in our search engine.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Intu- itively, layout is a universal and critical dimension for the quality assessment of different categories of webpages.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Based on this, we directly employ the meta-data that describes a webpage, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', Doc- ument Object Model (DOM) tree, as the input of our model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The DOM tree data unifies the representation of webpages with different categories and purposes and indicates the layout of webpages.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To assess webpage quality from complex DOM tree data, we propose a graph neural network (GNN) based method that extracts rich layout-aware information that implies webpage quality in an end- to-end manner.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Moreover, we improve the GNN method with an attentive readout function, external web categories and a category- aware sampling method.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We conduct rigorous offline and online experiments to show that our proposed solution is effective in real search engines, improving the overall usability and user experience.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' KEYWORDS Webpage Quality Models, Graph Neural Network, Information Re- trieval, Search ACM Reference Format: Anfeng Cheng∗, Yiding Liu∗, Weibin Li∗, Qian Dong, Shuaiqiang Wang, Zhengjie Huang, Shikun Feng, Zhicong Cheng and Dawei Yin§.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2023.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Layout- aware Webpage Quality Assessment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In SIGKDD ’23: ACM Special Interest Group on Knowledge Discovery and Data Mining, August 06-10, 2023, Long Beach, CA.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' ACM, New York, NY, USA, 11 pages.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' https://doi.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='org/XXXXXXX.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' XXXXXXX ∗ Co-first authors.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' § Dawei Yin is the corresponding author.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Copyrights for components of this work owned by others than ACM must be honored.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Abstracting with credit is permitted.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Request permissions from permissions@acm.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='org.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' SIGKDD ’23, August 06–10, 2023, Long Beach, CA © 2023 Association for Computing Machinery.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' ACM ISBN 978-1-4503-XXXX-X/18/06.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='..' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='$15.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='00 https://doi.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='org/XXXXXXX.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='XXXXXXX 1 INTRODUCTION Search engines, such as Google and Baidu, plays an important role in fulfilling users’ information need.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Over the past decades, relevance modeling is the main concern of search engines, dedicated to putting the most relevant web content on top of the ranked results [23, 28, 38, 42].' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' However, the very fact that not all relevant contents are useful to users has become an increasingly serious symptom, where relevant webpages with low quality would induce a significant cognitive burden on the user.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In such cases, useful information is hard to be identified, and the users need to take extra effort to understand the information to be conveyed.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To reduce the cognitive burden, measuring the quality of webpages has become a critical concern, which can better benefit users with well-delivered information and improve the overall usability of a search engine.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' For example, given webpages with comparable relevance, high-quality webpages should be ranked higher than its competitors.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Nevertheless, webpage quality assessment is a very important yet challenging task in web search, due to the complexity and diversity of webpages in the era of web 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' A webpage with a clear structure, tidy organization and concentrated delivery of crucial information is always preferable to one that only stacks content without proper presentation, even though each may contain similar information.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Therefore, an accurate assessment of webpage quality can facilitate a search engine to reduce the cognitive burden and more effectively provide useful information for users.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Previous attempts at webpage quality assessment mainly aim to manually design discriminative features [5, 14, 16, 24], where classification algorithms [7, 14] are applied subsequently.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' How- ever, modern search engines usually face trillions of webpages with various categories, where simple hand-crafted features and classification algorithms (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Bayesian Networks [7]) can hardly capture the in-depth information that reveals the webpage quality.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Moreover, most of them can only work on a particular category of webpages, e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', shopping websites [8], medical website [25], web portals [6], and Wikipedia articles [17].' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' They are hard to be effec- tively applied to real-world search engines that serve trillions of heterogeneous webpages.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To address the aforementioned limitations, we conduct the first work that investigates layout-aware webpage quality assessment on real-world web data.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The intuition is based on the findings that the quality of a webpage is largely determined by its content layout [6, 24], which is of a great influence on how users perceive textual and multi-modal content [20, 31, 34, 35, 41].' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Modeling in-depth layout information is promising for webpage quality assessment in real- world web search scenario.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' However, it is also very challenging, where two crucial research questions need to be answered: RQ1: How to capture the layout information of different categories of webpages in a unified manner?' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' arXiv:2301.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='12152v1 [cs.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='IR] 28 Jan 2023 SIGKDD ’23, August 06–10, 2023, Long Beach, CA Cheng and Liu, et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' <div> <p> <img> <html> <head> <meta name="viewport" content="width=device- width,initial-scale=1"> <link href=" style.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='css" rel=" stylesheet”> <title>Critical Path

Hello web performance students!' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='

Figure 1: A toy example that shows the webpage layout represented by the DOM tree.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' RQ2: How to encode webpage layout information for webpage quality assessment on large-scale heterogeneous webpages?' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To answer RQ1, we propose to extract layout information of webpages from Document Object Model (DOM) data.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Specifically, DOM is a cross-platform and language-independent interface that treats a webpage as a tree structure wherein each node is an object representing a content piece of the webpage.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Figure 1 shows a toy example of the hierarchical structure of a DOM tree, which is converted from its HTML source code.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Each node in the tree is an object that contains partial content of the webpage and is associated with different attributes that describe the content (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', type and size).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Such data indicates rich hierarchical in- formation on the content and its layout.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' And different categories of webpages can be represented in a unified manner.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' It is indisputable that inspecting the structure of DOM tree can help to measure the quality.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' For RQ2, it is very challenging as webpages in real search en- gines are highly diverse, where the modeling of layout information should be expressive to reveal the underlying patterns of hetero- geneous DOM tree data.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Recent advances in deep representation learning [3, 22] have achieved great success on many web applica- tions [9, 13, 28], and also sheds new light on our task at hand.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Notably, Graph Neural Networks [12, 15] has shown great per- formance in modeling structured text (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', word interactions) [18, 37, 41], yet they are unexplored for complex DOM tree structure.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Different from structure of textual document, the webpage layout represented by DOM tree is more complicated, which usually has hierarchical structure and the nodes usually have rich attributes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Existing methods that designed for text structure usually lack spe- cialized consideration for the problem of quality assessment on DOM tree data, and thus might be unsatisfactory for real search engines.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To this end, we propose the first GNN-based method to learn the underlying semantics of webpage layout in an end-to-end manner, based on which we further make several improvements to advance its performance on the task of webpage quality assessment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To verify the effectiveness of our layout-aware webpage quality assessment model, we perform offline experiments on the dataset collected by the real-world search engine.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Additionally, we deployed our model in the online ranking system and achieve good improve- ments.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Last but not least, the proposed solution is currently fully deployed in the online system of Baidu Search.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To illustrate how layout-aware webpage quality assessment facilitates the overall usability of our search engine, we further present the details of the model deployment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Overall, our main contributions can be summarized as follows.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We develop the largest application of deep learning for the problem of webpage quality assessment, which significantly improves the overall usability of real-world search engines.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We leverage DOM tree data and propose a GNN-based so- lution to learn the quality information of heterogeneous webpages in an end-to-end fashion.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We present the deployment of webpage quality assessment model in the real production environment, which effectively serves trillions of webpages with various categories and purposes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We conduct rigorous offline and online experiments before fully deploying the model online.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The experimental results show that the proposed solution is effective to be applied in real-world search engines.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2 RELATED WORK 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='1 Graph Neural Network Recent years have witnessed the success of graph neural networks (GNNs) for relational data.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' For example, Graph Convolutional Net- work (GCN) [21] is introduced to aggregate the one-hop neigh- bours of each node in the graph, followed by a linear projection and non-linear activation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' GraphSAGE [15] is proposed to generalize GCN’s aggregation operation from average to sum, max and a RNN unit.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Graph Attention Network (GAT) [30] employs the attention mechanism into GNNs, which allows GAT to assign different im- portance to nodes within the same neighbourhood.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Generally, a GNN can be regarded as using the input graph structure as the com- putation graph for message passing [12], during which the local neighbourhood information is aggregated to get a more contextual representation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' For more details, please refer to [32].' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Moreover, there are many applications across various domains that apply GNNs and achieve considerable improvements, such as protein model quality assessment [2, 26], fuel ignition quality assessment [27], advertising detection [36] and text classification [11, 19, 37, 41].' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' For example, Sanyal et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [26] explore an alterna- tive approach and train a graph convolutional network with nodes representing protein atoms and edges connecting spatially adja- cent atom pairs.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' GraphQA [2] is a graph-based method to estimate the quality of protein models, that possesses favorable properties such as representation learning.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Schweidtmann et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [27] develop GNN models for predicting three fuel ignition quality indicators of oxygenated and non-oxygenated hydrocarbons.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Yang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [36] propose WTAGRAPH, a web tracking and advertising detection framework based on graph neural networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Notably, a handful of researches [18, 37, 41] leverage GNN to perceive text structure (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', word interactions) for down-stream tasks, which are the closest research to our study.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' However, they mainly consider the relationships between segments of text (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', words and paragraphs), and do not consider the overall structure and layout of webpage, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', how the multi-modal content is orga- nized and presented.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In addition, the expressiveness of GNN is not explored for the task of webpage quality assessment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In this paper, we develop the first GNN-based method for webpage quality assess- ment, which is further deployed in the real production environment that facilitate the usability of our search engine.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Layout-aware Webpage Quality Assessment SIGKDD ’23, August 06–10, 2023, Long Beach, CA 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='2 Layout Layout, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', how the contents are organized and presented, is a critical dimension for document generation [4], scene recognition [1, 10] and webpage quality assessment [16, 24].' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Substantial efforts have been made to explore the importance of layout in many AI areas.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' For example, Besides, Biswas et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [4] design an automated deep generative model using graph neural networks to generate synthetic data with highly variable and plausible document layouts.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Avetisyan et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [1] use a message-passing graph neural network to model the inter-relationships between objects and layout, guiding the generation of a global object alignment in a scene.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Chen et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [10] build a Layout Graph Network (LGN) where regions in PaSL are defined as nodes and two kinds of independent relations between regions are encoded as edges.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Zhang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [40] propose a new kind of classification method for lithography layout patterns based on graph convolution network.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Recently, with the development of pre-trained language models, LayoutLM-style methods have achieved success in textual or multi- model document understanding.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' For instance, LAMPreT [31] en- codes each block with a multimodal transformer in the lower-level, and aggregates the block-level representations and connections utilizing a specifically designed transformer at the higher-level.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Lay- outLM [34] jointly model interactions between text and layout in- formation across scanned document images, which is beneficial for a great number of real-world document image understanding tasks such as information extraction from scanned documents.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' LayoutLM- v2 [35] further uses the new text-image alignment and text-image matching tasks and integrates a spatial-aware self-attention mech- anism into the Transformer architecture.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' LayoutLM-v3 [20] pre- trains multi-modal Transformers for Document AI with unified text and image masking.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Our work differs from the aforementioned studies, as our main focus is to model the layout of webpages for quality assessment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In particular, we model the layout of webpage with a graph con- struction method that represents a DOM tree as a layout graph and employ an expressive graph neural network to capture the underly- ing semantics of the layout graphs for webpage quality assessment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 3 PRELIMINARIES In this section, we introduce the basic concepts and formalize the problem of webpage quality assessment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We summarize the com- monly used notations in Table 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='1 Layout-aware Webpage Quality Intuitively, high-quality webpages are those that clearly provide useful information for users in common.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Specifically, given a set of webpages with comparable relevance under the same query, we consider the layout (i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' structure design, content presentation) as the key dimension of measuring webpage quality and improving user experience [6, 24].' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Based on this, we can construct a set of rules and principles for annotating webpage quality and utilize human annotation as the objective of our proposed method.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The considering aspects of rules and principles to score the layout of a webpage are shown in table 2, including interactive Experience, paragraph and layout design.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We give the definition Table 1: Commonly-used notations.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Notations Descriptions G𝑝 = {N,' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' E} ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='A layout graph of webpage 𝑝 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='N ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The node set of graph G𝑝 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='E ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The edge set of graph G𝑝 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='F = {F𝑡 }𝑇 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑡=1 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The layout-related feature sets ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑇 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The number of node type ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='F𝑡 = {f𝑖}|F𝑡 | ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑖=1 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The feature set of node type 𝑡 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='f𝑖 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The layout-related feature ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='E(·) ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The embedding of its input ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='�ℎ(0) ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑛 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The initialized embedding of node 𝑛 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑡𝑛 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The node type of node 𝑛 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝜷𝑝 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The category of webpage 𝑝 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='�ℎ(0) ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑣 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The initialized embedding of virtual node 𝑛 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝜎(·) ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='An activation function ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝛼𝑛𝑚 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The attention score between nodes 𝑚 and 𝑛 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑒𝑛𝑗 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The attention coefficient between nodes 𝑚 and 𝑗 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑠𝑝 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The predicted assessment score of webpage 𝑝 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑦𝑝 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='The manually assessment score of webpage 𝑝 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='and some examples for each aspect.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The rules and principles for annotators are defined as the following: 0 means poor layout.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' On the basis of ordinary pages, points will be deducted for various flaws.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 1 means ordinary layout.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 1 point is common, and annotators are required to add or deduct on this basis.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='5 means better structure.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' a certain gain compared to ordi- nary layout.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2 means gainful layout.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The user experience of this layout is significantly better than most layouts.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Based on these principle, bonus and deduction rules can be for- mulated as: webpages with reasonable & beautiful layout or rich information will have an extra bonus, on the contrary, unreason- able & chaotic layout or valueless information will be deducted.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Finally, annotators are required to score the give webpage from 0 to 2 points based on the above rules and principles.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='2 Layout Graph To extract quality information from webpage layout, we construct a layout graph for each webpage based on its DOM tree.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In particular, a layout graph is denoted as G𝑝 = {N, E} that contains a node set N and an edge set E.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Each node has a specific type (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', text, image and video), and is associated with several layout-related features F𝑡 = {f𝑖}|F𝑡 | 𝑖=1 .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The features of different types of nodes are denoted as F = {F𝑡 }𝑇 𝑡=1, where𝑇 is the total number of node types.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Besides, each layout graph is also associated with the category of SIGKDD ’23, August 06–10, 2023, Long Beach, CA Cheng and Liu, et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Table 2: The considering aspects of rules and principles to score the layout of a webpage.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Aspects Definition Examples Interactive Experience Whether the webpage has interactive function Click to call, swipe to browse pictures Paragraph Ways to split the document into paragraphs Using different heading, special font color to layering Layout Design The overall design of the webpage’s layout many additional functions, various modules, font section size is appropriate its webpage, which is denoted as 𝜷𝑝.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The detailed construction process of layout graph is depicted in Section 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='3 Webpage Quality Assessment Given a layout graph G𝑝 = {N, E}, its features F = {F𝑡 }𝑇 𝑡=1 cate- gory 𝜷 and F𝑡 = {f𝑖}|F𝑡 | 𝑖=1 , the task of webpage quality assessment is to estimate a score 𝑠𝑝 for a given webpage 𝑝 w.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='r.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='t.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' its quality, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e, 𝑠𝑝 = 𝑓𝜃 (G𝑝, F𝑡, 𝜷𝑝), (1) where 𝑓 (·) represents the quality model, and 𝜃 denotes its param- eters.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The scores should be consistent with users’ perception of webpage quality, and reflect the rules and principles as we described above.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 4 METHOD In this section, we first present the overview of our model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Then, we describe the graph formulation process for a webpage, including the construction of the layout graph and feature pre-processing.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' After that, we present a GNN-based solution for webpage quality assessment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='1 Overview Our solution mainly contains two components: layout graph for- mulation (i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', Section 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='2) and quality assessment model (i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', Sec- tion 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='3).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In layout graph formulation, we first leverage the layout information encoded in DOM tree to construct a layout graph G𝑝 = {N, E} for every webpage 𝑝.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Then, two types of features are designed for the quality assessment, as depicted in Figure 2, includ- ing those associated with each node in the graph, as well as the category of the corresponding webpage (i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', 𝜷𝑝).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Next, we propose a quality assessment model that leverages Graph Attention Network (GAT) to perform expressive message passing between nodes in the layout graph.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Both local and global structure information of the layout graph can be encoded in latent representations, which are exploited for the quality assessment task.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Moreover, we improve the vanilla GAT model by 1) introducing an attentive readout function via the virtual node, 2) incorporating graph-level category information in the scoring function, and 3) alleviating the data imbalance problem that is common in real-world applications.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='2 Layout Graph Formulation Graph construction.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The content layout has been viewed as one of the most critical dimensions for measuring webpage qual- ity [24].' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To formulate layout information for various categories of webpages, we first construct layout graph based on DOM tree.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In Table 3: The summary of selected features for each node type in our constructed graph.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Classification Feature Name Location height, width, xpos, ypos, position type Content number of word, font size, font style, line height, font weight, alignment Layout border, padding, margin, visibility, display style, outline style, outline width Others tag name, webpage category particular, we leverage HTML parser Beautiful Soup 1 to parse the source code of a webpage, identifying the hierarchical structure of the webpage.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Then, Depth First Search (DFS) is used for exact- ing adjacency relationships from the DOM tree.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Specifically, we recursively record the nodes and the corresponding edges between parent and child nodes in the DOM tree, as shown in Algorithm 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The layout graph G𝑝 of webpage 𝑝 can be expressed by the exacted nodes N and their relations E.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Virtual node.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' It is worth noting that, we also include a global virtual node that connects to all the other nodes in the graph (as shown in Figure 2).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' It can be viewed as a super-hub [39] of the layout graph, which could be useful to aggregate the global information, and serves as hyperlinks that connect any two nodes in the layout graph.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' As such, we can capture global information of the given graph via the virtual node.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Feature pre-processing.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To capture the layout information of the webpage, we design a series of features for each node type.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Taking the text node as an example, font style, font size, alignment and position in webpage are all represented by learnable embedding.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The detailed list of features is presented in Table 3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' More specifically, for continuous features (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', height, line height and margin), a non-uniform interval division strategy is employed to divide the continuous interval into several buckets, which can ensure that there are enough training samples in a single bucket.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The uniform division of the whole interval leads to the data sparse issue since the continuous features typically obey a long-tail dis- tribution.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Discrete features (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', font style, display style and tag name), are falling on a divided interval are mapped into a corre- sponding bucket, and this bucket is assigned a learnable embedding to represent the characteristics of its interval.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 1We parse webpages with the python library: https://www.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='crummy.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='com/software/ BeautifulSoup/bs4/doc/ Layout-aware Webpage Quality Assessment SIGKDD ’23, August 06–10, 2023, Long Beach, CA Layout-Related Features .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='..' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Virtual Node Layout Graph Node Features Graph Features Projection Page Score Figure 2: The illustration of message passing in our model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The red node represents the virtual node in the constructed layout graph, which is utilized for capture the graph-level information.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Algorithm 1: Layout Graph Construction Input: HTML DOM tree R𝑝 of webpage 𝑝 Output: layout graph G𝑝 = {N, E} of webpage 𝑝 % recursive graph construction;' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' GraphConstruction(root) begin N𝑟 = {𝑣𝑖𝑟𝑡𝑢𝑎𝑙_𝑛𝑜𝑑𝑒, root.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑛𝑜𝑑𝑒};' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' E𝑟 = {(𝑣𝑖𝑟𝑡𝑢𝑎𝑙_𝑛𝑜𝑑𝑒, root.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑛𝑜𝑑𝑒)};' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' for child ∈ root.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑐ℎ𝑖𝑙𝑑𝑟𝑒𝑛 do N𝑐 = child.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑛𝑜𝑑𝑒;' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' N𝑟 = N𝑟 ∪ N𝑐;' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' E𝑐 = {(𝑣𝑖𝑟𝑡𝑢𝑎𝑙_𝑛𝑜𝑑𝑒, child.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑛𝑜𝑑𝑒), (child.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑛𝑜𝑑𝑒, root.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='𝑛𝑜𝑑𝑒)};' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' E𝑟 = E𝑟 ∪ E𝑐;' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' N𝑐, E𝑐 = GraphConstruction(child);' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' end return N𝑟, E𝑟 end G𝑝 = {GraphConstruction(Rp)};' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In addition to the node-level features, graph-level feature em- bedding is introduced to the layout graph (i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', webpage category) to provide the model with the ability to perceive different cate- gories of webpages, which is vital to the quality assessment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' One reason is that the same webpage category has a similar structure.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' With the development of webpage makers (like Dreamweaver, and Google Web Designer), large amounts of webpages are generated from templates and almost in the same layout.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Therefore, with this graph-level embedding, the predicted assessment score shall be more robust in the online search engine.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Another reason lies in that different webpage categories have different criteria for quality assessment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' For example, a succinct and well-organized document layout without distracting pictures is preferred on a search page, but for a portal, a document layout with pictures and text is con- sidered to be better.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In summary, it is meaningful and important to take the graph-level feature embedding into account for the layout graph.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='3 Quality Assessment Model Given the constructed layout graph associated with rich features, the key of webpage quality assessment is to expressively reveal salient patterns underlying the graph.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In particular, we consider two types of relationships in the graph that could be discriminative for the task: Local relationships.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Intuitively, the relationships between adjacent nodes in the layout graph are important to reveal content quality.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' For example, a node with tag is usu- ally the illustration of its adjacent (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', parent) node with
tag, which contains textual description.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The interac- tion of the two nodes indicates the web content has both visual and textual presentation, forming a strong signal of high-quality content.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Global relationships.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Another important insight is that the relationships between local content and global layout should also be considered.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' For example, a node with textual description might be critical in a news article but is less important in a video webpage, whose quality largely depends on the node that contains the video.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Attentive message passing.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To achieve this, we leverage graph neural networks that are promising to capture such complicated patterns.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In particular, we utilize the Graph Attention Network (GAT) [30] to model the interactions between nodes in the layout graph, where the modeling of node relationships can be viewed as message passing [12] among nodes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In particular, the architecture of GAT is composed by stacking multiple graph attention layers, each of which can be defined as �ℎ(𝑘+1) 𝑛 = 𝜎 �� � ∑︁ 𝑚∈N𝑛 𝛼𝑛𝑚W(𝑘) 1 �ℎ(𝑘) 𝑚 �� � , (2) where 𝜎(·) is an activation function and 𝛼𝑛𝑚 is the attention value between node 𝑛 and node 𝑚.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Here, �ℎ(𝑘) 𝑛 represents the embedding of node 𝑛 in the 𝑘-th layer.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The attention value 𝛼𝑛𝑚 is learned to selectively propagate information from neighbour node 𝑚 to node 𝑛, and a node can attentively interact more with its important neighbours than those trivial ones.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Formally, the attention value SIGKDD ’23, August 06–10, 2023, Long Beach, CA Cheng and Liu, et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' can be defined as 𝛼𝑛𝑚 = softmax𝑚 (𝑒𝑛𝑚) = exp (𝑒𝑛𝑚) � 𝑗 ∈N𝑛 exp �𝑒𝑛𝑗 � , (3) where the logits 𝑒𝑛𝑗 is computed as 𝑒𝑛𝑗 = 𝜎 � W(𝑘) 3 [W(𝑘) 2 �ℎ(𝑘) 𝑛 ∥W(𝑘) 2 �ℎ(𝑘) 𝑗 ] � .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' (4) Here, we use ∥ to represent the concatenation operation, and W(𝑘) 2 and W(𝑘) 3 are the weight matrices of the linear transformations at the 𝑘-th layer.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Note that the weight matrices are shared across different nodes in a single graph attention layer.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' After 𝐾 times of message passing, the layout-aware patterns could be captured by node interactions (as defined in Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' (2)) within 𝐾-hops.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' It is worth noting that the virtual node also plays an im- portant role during the message passing process.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The virtual node offers a pathway for nodes’ interaction with considering the global interactions in the graph, which is critical for the quality assess- ment task.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Overall, the GAT-based message passing framework is able to comprehensively model both local and global relationships for the final task.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Readout function.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To compute the final quality score, we de- fine the readout function as mean-pooling [15, 32] to summarize all node representations as the final graph representation, and sub- sequently adopt a linear layer as 𝑠𝑝 = W mean_pooling( �𝐻 (𝐾) N ) + 𝑏, (5) where �𝐻 (𝐾) N is the set of node representations in 𝐾-th layer of GAT.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Alternatively, we can apply a more reasonable readout function, which is to use the representation of the virtual node as the final graph representation, and rewrite Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' (5) as 𝑠𝑝 = W�ℎ(𝐾) 𝑣 + 𝑏, (6) where �ℎ(𝐾) 𝑣 is the virtual node representation in 𝐾-th layer (i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' the last layer) of the model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In such case, the aggregation on the virtual node can be viewed as an attentive readout function, which has the capability of distinguishing the impact of different nodes in the graph for the final task.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Category-aware quality assessment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The quality score defined in Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' (6) is based on rich information aggregated from nodes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' How- ever, graph-level information is critical yet not incorporated.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' There- fore, we further improve Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' (6) with the category information of webpage.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In particular, we denote the category embedding of a given webpage 𝑝 as E(𝜷𝑝), and further rewrite Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' (6) as 𝑠𝑝 = W(�ℎ(𝐾) 𝑣 + E(𝜷𝑝)) + 𝑏.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' (7) Note that the category embedding E(𝜷𝑝) has the same dimension- ality as the graph embedding �ℎ(𝐾) 𝑣 , such that the embeddings could be summed for the final assessment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Category-aware data sampling.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' As the graph-level category embedding is introduced in Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' (7) to perceive different categories of webpages, the bias in different categories may affect the predic- tion of models.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In particular, some webpages are highly similar in layout, such as some popular question-answering websites, which are generated from templates.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Such webpages typically have sim- ilar layout scores.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Consequently, the predicted assessment score may be dominated by the category-aware embedding (i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' graph level embedding).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To alleviate this issue, a category-aware sampling strategy is employed.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Up-sampling is utilized to balance the number of two classes, based on which the bias could be mitigated and our model could learn a distinguishable quality assessment score for a single category of webpages.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Optimization objective.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' After up-sampling, the model could be optimized through Mean Squared Error (MSE) loss.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' It can be defined as 𝐽 = 1 𝑃 𝑃 ∑︁ 𝑝=1 �𝑦𝑝 − 𝑠𝑝 �2 , (8) where 𝑃 is the total number of training samples after up-sampling and 𝑦𝑝 is the annotated layout score of webpage 𝑝.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 5 DEPLOYMENT In this section, we show how the layout-aware webpage quality assessment model be applied to our online ranking system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We first introduce the input data construction process of the quality assessment model and then present the general picture of the quality score working in the ranking system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The overview of deployment is shown in Figure 3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='1 Offline Input Data Construction In the left component of Figure 3, we present the process of input data construction for our model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Firstly, each webpage on the world wide web will be parsed through our HTML parser.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' All features of the HTML are stored in a database.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Secondly, we construct the layout graph based on DOM tree and extract the features needed for quality assessment model using the algorithm defined in Al- gorithm 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Note that this process runs offline, it can significantly reduce the computing time of the online search system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We also list the features which are used in our webpage quality assessment model, details are shown in Table 3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We classify the features into three main categories w.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='r.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='t.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', location, content, and layout according to the different roles they play in building webpage.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Category location is the primarily feature that locates the position of elements in the webpage e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', height, width and position type.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Category content contains text-related features e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', the number of words, font style, and line height.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Category layout is a feature that controls the layout of elements, e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', border, padding, and margin.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In addition, we add tag name, natural categorical information, and webpage category, which is used to balance the distribution of train data under different webpage forms.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='2 Online System Workflow The online system workflow is presented in the right component of Figure 3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Our ranking system contains a wide variety of webpage features, where quality is one of the most important factors.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To apply our layout-aware webpage quality assessment model in our online retrieval system, the new quality scores need to be loaded into the retrieval feature list.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The online ranking system only needs to load the new quality assessment score and apply it to obtain the new ranking results with respect to the new ranking webpage list, Layout-aware Webpage Quality Assessment SIGKDD ’23, August 06–10, 2023, Long Beach, CA HTML Parser HTML Database Webpages Layout Graph Construction Input Data Database Webpage DOM Tree with Features Input Data of Model Virtual Node 𝑵𝟏 𝑵𝟐 .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='..' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' height width paddling margin font border .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='..' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Features image AirTag head html body title iMac p div height、width、margin、 border、padding、font size、font style、xpos、 content length、ypos、 overflow、visibility .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='..' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='..' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='. Layout-aware Quality Assessment Model Quality Score Database Online Ranking System new feature ranking list of each webpage .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='..' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='..' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='. Ranking System new ranking results Input Data Construction Online System Workflow Figure 3: The overview of deployment in online ranking system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' which is shown in the lower left area of the online component.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Note that, the quality assessment scores of all webpages are calculated offline and are independent of the online search query, thus are inefficient for the online search query.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 6 OFFLINE EVALUATION In this section, we conduct an offline evaluation of the proposed layout-aware webpage quality assessment model on the manually- labeled dataset from the search engine serves through the offline experiments.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 6.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='1 Dataset To evaluate the proposed method, we first collect a set of webpages from our database, which stores the real webpages that our search engine serves.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Next, we manually label all the collected webpages on our crowdsourcing platform, where a group of experts are required to assign low-quality (0) or high-quality (1) to each of the given webpage.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In our experiments, we use 600,000 webpages for training and 20,000 webpages for testing.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 6.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='2 Evaluation Metrics Positive-Negative Ratio (PNR).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We use PNR to measure the con- sistency between manual quality labels and the scores estimated by the model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In particular, by enumerating all the pairs of webpages in the dataset (i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', 𝐷), PNR can be formally defined as 𝑃𝑁𝑅 = � 𝑑𝑖,𝑑𝑗 ∈𝐷 I �𝑦𝑖 > 𝑦𝑗 � · I �𝑓 (𝑑𝑖) > 𝑓 �𝑑𝑗 �� � 𝑑𝑖′,𝑑𝑗′ ∈𝐷 I �𝑦𝑖′ > 𝑦𝑗′� · I �𝑓 (𝑑𝑖′) < 𝑓 �𝑑𝑗′�� , (9) where I is an indicator function, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', I (𝑎 > 𝑏) = 1, if 𝑎 > 𝑏, and 0 otherwise.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Here, 𝑓 (𝑑𝑖) represents the quality score of a webpage 𝑑𝑖 estimated by the model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Higher PNR value indicates better perfor- mance of the model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Area Under Curve, Precision, Recall, F1-Score.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We also report Area Under Curve (AUC), Precision (P), Recall (R) and F1-Score (F1) to evaluate our proposed model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Precision and recall are often in tension, that is, improving precision typically reduces recall and vice versa.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' F1-Score combines them to one performance metric.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Area under curve summarizes the trade-off between the true positive rate and false positive rate for a predictive model using different probability thresholds.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 6.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='3 Compared Baselines and Our Approach To validate the effectiveness of our layout-aware webpage quality model, we conduct experiments on several related baseline mod- els: TreeLSTM [29], a standard LSTM architecture designed for tree-structured network topologies.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' GIN [33] introduces a learnable parameter to adjust the weight of the central node.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' GAT [30] lever- ages the attention mechanism to improve neighbor aggregation scheme.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Our proposed models: Virt-GIN has a more expressive readout mechanism by adding the virtual node �ℎ𝑣 to GIN model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Virt-GAT is our approach similar to virt-GIN model, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', a GAT model with virtual node.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Models-NC: Note that all the above- mentioned models use category information as proposed in Section 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' To further clarify the influence of category in the model, we also include four variants without using category information, which is denoted with a suffix Non-Category (-NC).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In addition, we also compare our proposed method with Online Baseline, which is the quality assessment model that was previ- ously served online in our search engine.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' This can clearly illustrate the improvement brought by the proposed solution for our search engine.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 6.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='4 Experimental Settings In our experiments, Adam is selected as the optimizer.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We use the following hyper-parameters: embedding size (64), number layers (5), dropout probability (0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='2), batch size (32), learning rate (0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='0001) for GNN models, train epochs (25).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' As for the TreeLSTM model, we set the embedding size (64), dropout probability (0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='5), batch size (128), learning rate (0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='0001), epochs (25) for it.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We run 5 experiments with different random seeds for all models mentioned above.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The <广告> iMac 新开篇 进一步了解》 购买> AirTag 丢三落四这门绝技，要失传了。' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 进一步了解> 购买>米非可酮片购买 QQ: 你可能还想找： 吃家非司限片有什么及应 吃家非司限片有什么别作用 吃家非司职片会出自码 第一天吃来非可期片 晚来非脲片 晚完来非司片的反度 晚来非司限片有什么用 惊 咨询药师 吧药师微信用品益新技检影音乐/安全用品/电子电型改装用 品/外维用品内信用品/养护用品自然范用品工投 检 品牌特区：车墙土 送进佳 Z室组调，今天小学生网小编竭据老师给大家整理了 关于汉字（室》的绳调列表，基望下西整理的竞字 组调资科及调语解择内容能够助到大家， 室字简介 首字母：y，群膏：yu，等声调拼音：yo，注音： U，部首：穴，部首比划：5，比划：15，第体 字：毫，字体结构：上下结构，第画顺序：擦擦折 PWRY，五第98编码: PWRY， Unicode : 服擦操推所除择服摄折探除，五笔86编码 U+7AB3，双字编号：6008, 基本解释 宝yo Uo（事物）思务，租务：室务，室 败（房效；' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='数坏），室陷（雅务，质量根差），良 室（优务），0量：室情，0蜜第 京组调 掌室(beny）：掌重相劣，清线源（圣式记》 卷/：“面官修战股，零意不能放洋，转座高组力 剩摄之用。' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 事室(bbye)：(1).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='泄气；' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='干事，如：气球欢得个 头抵大，但用针一别就癌富了，(2).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='坑，童度，杨 《麦子黄时》：“自卫队上操，有时练习石 锁，他能单手掌置负子，一口气连孕十几下，后一 敬手，稳出七八步运，肥场地打个大靠，SIGKDD ’23, August 06–10, 2023, Long Beach, CA Cheng and Liu, et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Table 4: Offline experimental results of different models.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Model PNR AUC (%) label 0 label 1 P (%) R (%) F1 (%) P (%) R (%) F1 (%) Online Baseline 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='51 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='10 73.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='44 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='99 68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='39 40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='09 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='57 47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='60 TreeLSTM 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='91 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='01 74.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='93 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='07 79.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='69 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='01 81.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='86 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='07 80.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='76 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='03 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='64 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='06 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='19 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='06 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='86 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='05 GIN 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='27 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='05 81.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='26 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='20 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='82 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='39 81.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='04 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='17 82.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='40 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='44 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='22 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='96 65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='64 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='45 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='34 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='26 GAT 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='43 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='06 81.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='94 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='23 84.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='87 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='03 79.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='93 ± 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='21 82.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='30 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='69 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='00 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='52 68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='65 ± 3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='40 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='53 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='75 Our Approach Virt-GIN 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='62 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='03 82.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='47 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='10 85.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='23 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='41 78.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='95 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='49 81.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='96 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='61 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='26 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='20 69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='95 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='56 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='72 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='14 Virt-GAT 5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='22 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='10 84.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='18 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='24 86.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='81 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='57 80.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='17 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='13 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='35 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='35 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='75 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='79 73.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='24 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='71 67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='57 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='29 Non-Category (-NC) GIN-NC 4,15 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='07 80.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='80 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='30 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='36 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='42 81.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='29 ± 3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='57 82.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='25 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='19 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='24 ± 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='62 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='20 ± 5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='15 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='50 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='17 GAT-NC 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='26 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='04 81.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='27 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='15 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='85 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='23 81.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='11 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='75 82.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='46 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='31 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='32 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='67 65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='70 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='87 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='43 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='25 Virt-GIN-NC 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='48 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='04 82.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='05 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='14 84.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='70 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='53 79.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='49 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='27 82.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='01 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='44 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='35 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='86 68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='45 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='77 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='13 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='33 Virt-GAT-NC 5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='03 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='03 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='66 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='08 85.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='99 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='48 81.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='40 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='23 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='62 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='43 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='47 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='04 70.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='86 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='60 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='94 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='25 final result we reported is the mean test AUC, Precision, Recall, F1-Score and their corresponding standard deviation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' All the above mentioned GNN models are implemented by Paddle Graph Learning (PGL)1, an efficient and flexible graph learning framework.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 6.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='5 Offline Experimental Results We report the offline experimental results of the proposed model and all baseline models.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Besides, we also include a baseline method, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', the model that is used in the system before deploying the layout- aware webpage quality assessment model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' All results are shown in Table 4, from where we have the follow- ing key findings: We can clearly see that our layout-aware webpage qual- ity model can beat the online baseline by large margins on all metrics e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', Δ𝐴𝑈𝐶 = 24.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='08, Δ𝐹1 = 14.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='96 (label0) and Δ𝐹1 = 19.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='97 (label1).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Especially for PNR, where the value is improved from 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='51 to 5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='22.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' These tell us that the proposed model prefers high-quality results.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' By applying the proposed readout function, the model can have a significant improvement on all metrics.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Especially, the new readout mechanism is able to improve PNR by a margin of 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='38 and 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='96 based on GIN and GAT, respectively.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Moreover, we also observe that the relative improvement of both virt-GIN and virt-GAT over GIN and GAT is consid- erable for high-quality webpage (label1), in terms of recall (Δ(𝑉𝑖𝑟𝑡_𝐺𝐴𝑇,𝐺𝐴𝑇) = 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='59%, Δ(𝑉𝑖𝑟𝑡_𝐺𝐼𝑁,𝐺𝐼𝑁 ) = 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='22%).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' All these phenomena show that our readout mechanism is capa- ble of improving the model’s performance.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Comparing the results of the two models whether apply the category-aware optimization strategy (w,r,t.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', GIN-NC vs.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' GIN, Virt-GIN-NC vs.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Virt-GIN, GAT-NC vs.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' GAT, Virt- GAT-NC vs.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Virt-GAT), we can come to the conclusion that all methods with the proposed category-aware optimization have better performance than their backbone models, in terms of PNR and AUC.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Although a few models obtain lower 1https://github.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='com/PaddlePaddle/PGL values on a few metrics (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', the F1-score of Virt-GAT-NC on label0 is 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='62 while Virt-GAT is 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='35, the precision of Virt- GAT-NC is 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='74% but Virt-GAT is 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='75%), the models with category-aware optimization show more robust performance considering all metrics.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The performance on different GNN models is better than TreeLSTM, model Virt_GAT is the most significant, Com- pare with Virt_GAT and TreeLSTM, Δ𝑃𝑁𝑅 = 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='31, Δ𝐴𝑈𝐶 = 9.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='25%.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' For high-quality webpage (label1) Δ𝑅 = 14.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='67%.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' These large margins suggest that our model is more expressive than TreeLSTM, although TreeLSTM is specifically designed for tree-structured network topologies.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Overall, our proposed model is able to gain superior performance on webpage assessment task through the improved readout mech- anism and category-aware optimization and can beat the online baseline by a significant margin.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 6.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='6 Varying the number of GNN layer In general, a webpage is represented as a DOM tree.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Its depth deter- mines how many layers of GNN are needed to obtain information from the root node to the leaf nodes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' However, as the number of GNN layers increases, the computational efficiency will be lower.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Therefore, we provide an experiment to verify the influence of the number of layers on the experimental results, as shown in Table 5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' As seen from the table, the more layers, the higher the AUC score can be reached.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' However, compared with the 5-layer virt- GAT model, the improvement of 7-layer virt-GAT model is not significant.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' As it is important to trade off the efficiency and effec- tiveness for large search system, we use 5-layer GNN models on online evaluation which can maintain the experimental effect while reducing the amount of calculation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 7 ONLINE EVALUATION To investigate the impact of our proposed quality assessment model to the search engine, we deploy the new model and conduct online experiments to compare it with the old retrieval system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Specifically, Layout-aware Webpage Quality Assessment SIGKDD ’23, August 06–10, 2023, Long Beach, CA Table 5: The influence of layer number on virt-GAT.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' #Layers AUC (%) label 0 label 1 P (%) R (%) F1 (%) P (%) R (%) F1 (%) 1 80.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='77 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='23 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='38 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='92 80.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='46 ± 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='49 81.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='87 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='85 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='26 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='82 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='72 ± 3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='42 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='33 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='66 3 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='80 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='27 86.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='25 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='64 79.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='80 ± 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='06 82.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='89 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='86 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='98 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='79 72.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='05 ± 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='16 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='59 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='45 5 84.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='18 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='24 86.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='81 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='57 80.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='17 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='13 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='35 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='35 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='75 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='79 73.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='24 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='71 67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='57 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='29 7 84.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='25 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='22 86.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='91 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='80 80.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='53 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='77 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='58 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='61 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='23 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='41 73.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='32 ± 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='43 67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='86 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='42 we conduct a manual evaluation on the final ranking results with some real user-generated queries.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' This directly reflects the quality of the results exposed to the end users.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We log a set of (million-scale) online queries and the correspond- ing final impressions, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', the top-ranked web documents in the final ranking stage, by individually using the layout-aware web- page quality assessment model and the old retrieval systems.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Note that the data logging is conducted by multiple rounds to eliminate randomness.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We filter out examples in which queries have identical impressions between the two systems, and then utilize the rest for the manual evaluation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Note that, considering the extremely high cost of the manual evaluation, we randomly generate thousands of data and eventually send it to experts for evaluation, so as to control costs while validating the effectiveness of the proposed model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 7.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='1 Online Experimental Metrics As mentioned in Section 5, our proposed quality assessment model works in Baidu retrieval system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The online experiments major focus on the end-to-end evaluation, the metrics are often used to measure the effectiveness of information retrieval system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Details are as follows: Discounted Cumulative Gain (DCG).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We first log a dataset and manually label the data with 0 to 4 grades, and then report the relative improvement w.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='r.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='t.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' the average DCG over the top-4 final results of all queries.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The formula of DCG accumulated at a particular rank position p is defined as DCGp = 𝑝 ∑︁ 𝑖=1 2𝑟𝑒𝑙𝑖 − 1 log2(𝑖 + 1) , (10) where 𝑟𝑒𝑙𝑖 indicates the manually label of 𝑖-th webpage.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Additionally, we also report the relative improvement of DCG for the low quality ranking result w.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='r.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='t.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=', manually label is 0/1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Side-by-side Comparison.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Besides, we also conduct a side-by- side comparison between the two systems.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We log another dataset and require the human experts to judge whether the new system or the base system gives better results that satisfy intentions of users.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Here, the relative gain is measured Good vs.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Same vs.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Bad (GSB) as Δ𝐺𝑆𝐵 = #Good − #Bad #Good + #Same + #Bad, (11) where #Good (or #Bad) indicates the number of queries that the new system provides better (or worse) final results.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Table 6: Discounted cumulative gain on manual evaluation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Rand-Query Tail-Query Same-Quality Δ𝐷𝐶𝐺 +0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='19% +0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='42% DCG_0/1 ratio 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='63% 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='56% Table 7: Side-by-side comparison on manual evaluation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Rand-Query Tail-Query Same-Quality Δ𝐺𝑆𝐵 +4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='10% +0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='52% +5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='13% Node that we not only measure the final results but also measure the webpage quality when the relative result of two webpage is Same.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 7.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='2 Online Experimental Results The relative improvement validated by manual evaluation is given in Table 6 and 7, where we can summarize observations as below: By applying our quality assessment model, the system can significantly outperform the base system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Especially for DCG_0/1 ratio, the relative improvement values are respectively −0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='63%, −0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='56% for rand query and tail query.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' This shows that our proposed method can better filtrate retrieval results with low DCG scores, which is very helpful in improving the user experience for real-world search engine.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The conventional case-by-case comparison also has signifi- cant improvement over the base system, especially for the rand query (Δ𝐺𝑆𝐵 = +4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='1%).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' This tells us that user experi- ence can be improved by taking into account the web page quality in search system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In addition, we can observe that with comparable relevance, the GSB value of the quality improvement is Δ𝐺𝑆𝐵 = +5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='13%.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' This intuitively shows that our new system can provide higher quality search results based on the guaranteed rele- vance of search results.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Moreover, we perform the statistical test to estimate whether the experimental results is statistically significant.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The p-value of DCG rand and tail query are 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='0613 and 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='1276, respectively.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The p- value approximates the significance level that is set in our retrieval SIGKDD ’23, August 06–10, 2023, Long Beach, CA Cheng and Liu, et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' (a) Offline quality assessment (b) Online position changes Figure 4: The overview of case study.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' system, which can demonstrate that our experimental results are statistically significant.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Overall, the online experimental results show that our proposed layout-aware quality assessment model can effectively improve the performance of real-world ranking system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 8 CASE STUDY In this section, we present an illustration that includes the offline quality assessment score of webpage and online position changes of web pages.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' These typically cases are shown in Figure 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 8.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='1 Offline Quality Assessment In Figure 4(a), we present three webpages with different layout styles and their quality assessment scores.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The first webpage has a chaotic layout, elements in this web- page are unreasonable.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' It affects the user’s normal browsing and is very difficult for user to obtain information from this webpage.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Our quality assessment model marks this webpage as low quality (𝑠𝑐𝑜𝑟𝑒 = 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='0068).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' This extremely low score will be considered by the ranking system to lower its ranking position.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The second webpage also has low quality, different with the chaotic layout of the first webpage, it has a normal layout.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' How- ever, considering that it contains very small amount of information (almost no valuable information), it should be presented to the user with a very small probability.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' The ranking system can judge this by our quality assessment model score 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='1653.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Unlike the previous two webpages, the third one is high-quality.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' It is carefully laid out and informative, and quality score is 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='9788, which will help the ranking system raise its ranking position.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 8.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='2 Online Position Changes The case shown in Figure 4(b) comes from Section 7.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Under the same query, these two webpages swapped positions in the new and old systems, The position of the left webpage in new system is 3-th but 4- th in the old system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Comparing the two webpages, we can observe that the left webpage (quality score is 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='5623) contains a rich amount of information but the right one (quality score is 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='2415) does not.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' This phenomenon demonstrates that online ranking system has adopted our model’s recommendations to provide users with higher quality webpage, which can greatly improve the user experience.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 9 CONCLUSION AND FUTURE WORK In this paper, we propose a layout-aware webpage assessment model to suggest ranking system providing webpages with higher quality.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We not only enhance GAT with the read mechanism but also care- fully design the features for improving the quality assessment on the webpages.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In addition, taking into account the particularity of real-world data, we utilize the category of webpage for optimiza- tion.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Both input data construction and model calculation are offline, which guarantees the efficiency of the ranking system.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' We devel- oped and deployed the layout-aware webpage assessment model in Baidu Search, which is highly effective in conducting high-quality ranking for web search.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Extensive offline and online experiments have shown that the ranking system can significantly improve the effectiveness and general usability of the search engine.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In future work, we will explore the heterogeneous GNN architec- ture to model the multiple graph-based information of webpages.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' It is interesting to improve the construction method of layout and enhance the representation of nodes/edges with self-supervised contrastive pre-training techniques.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' REFERENCES [1] Armen Avetisyan, Tatiana Khanova, Christopher Bongsoo Choy, Denver Dash, Angela Dai, and Matthias Nießner.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' SceneCAD: Predicting Object Align- ments and Layouts in RGB-D Scans.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' ArXiv abs/2003.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='12622 (2020).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [2] Federico Baldassarre, David Ménendez Hurtado, Arne Elofsson, and Hossein Azizpour.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' GraphQA: protein model quality assessment using graph convo- lutional networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Bioinformatics 37 (2021), 360 – 366.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 安全国品电子电器政装用品 外饰用品内饰用品养护用品自空游用品正报价 M Opma适8轮胎防摄膜章道询 昆聘特区：车博土 巴洛克 喜普风 香玉儿 迪佳 键科 暖忆 H H H同酮片购买 Q@: 你可自能摄线： 吃米非司酮片后流血 吃完米非司酮片的反应 吃了米非司酮片流血 吃米非司酮片有���么症状 服米非司酮片 吃米非司酮片第二天出血 吃米非司酮片的注意事项 吃米非司酮片会流血吗 开门红 咨询药师 药师微信什公值得买 Q搜索分类/品牌/商品 打开 全部奶价 社区 商品百料 抗事等 要爽玩《魔兽世界：争霸艾泽拉斯》 CPU 鲁118-88-50 +美注 RYZEN WORLDL 于一 量 家宝业，额心后上靠玩安成的原国，可情不远 大本身视普世养料 MAGB550MMORTARWiFI迫击炮+.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='..' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' ￥2392 1899 抢 市场价 值得买APP专享价代公值得买 Q搜索分类/品牌/商品 打开 泡好价 全部好价 社区 商品百料 捷惠等 甲全部评论(119) 13心 3068-08-0 心 ALARE 心 3008488-30 爱有年天 4心 MAGB550MMORTARWiFI迫击炮+.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='..' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' X ¥2399 1899 抢 市场价 值得买APP专享价S搜狐网 晴/1°三 如何搭迪和自己不熟的女同事-如何搭让 如何搭训技巧之一：微笑地说出对方的名字 对于安生而言，如果一入男士非常绅士地对 她微笑，并具当着她的面，自然友好地叫出 了她的名字，她肯定会感到惊访，但随之而 来的更多是欢喜。' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='这种搭训会让女生瞬时记 住自己，并且留下较好的印象。' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 因为你微笑对她，她也会回以礼貌的微笑 然后她会反问：“你怎么知道我的名学”。' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='就 解中度过愉悦的时光。' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='彼此都给对方以缸服 的感觉，这为下次的聊天或相聚做好 垫。' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 讨论5女相100男，徐州一相亲大会男女比例失调.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='.> 男 女 我来说两句 C知乎 Q中国灵.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='. 下载App 注册登录 不认识的同事（女）如何搭训？' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 关注问题 写回答 1个回答 小白兔 谢邀，建议先从她认识的人入手，比如拿联系方 式，然后可以先切入聊天 发布于2020-11-1421:17 一费同 评论 智能消费 新浪潮 创卷调研 广告 相关推荐 男朋友打游戏正确的处理方式，这个 女朋友不能要了 微博的广告 如何去搭陌生人 haoyunlai2188的文章 ApP内打开Layout-aware Webpage Quality Assessment SIGKDD ’23, August 06–10, 2023, Long Beach, CA [3] Yoshua Bengio, Aaron Courville, and Pascal Vincent.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2013.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Representation learning: A review and new perspectives.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' IEEE transactions on pattern analysis and machine intelligence 35, 8 (2013), 1798–1828.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [4] Sanket Biswas, Pau Riba, Josep Llad’os, and Umapada Pal.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Graph-based Deep Generative Modelling for Document Layout Generation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In ICDAR Workshops.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [5] A Caro, Coral Calero, Ismael Caballero, and Mario Piattini.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2005.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Data qual- ity in web applications: A state of the art.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In IADIS International Conference WWW/Internet, Vol.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 364–368.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [6] Angélica Caro, Coral Calero, Ismael Caballero, and Mario Piattini.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2006.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Defining a data quality model for web portals.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In International Conference on Web Information Systems Engineering.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Springer, 363–374.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [7] Angelica Caro, Coral Calero, Houari A Sahraoui, and Mario Piattini.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2007.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' A Bayesian network to represent a data quality model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' International Journal of Information Quality 1, 3 (2007), 272–294.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [8] Selcuk Cebi.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2013.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' A quality evaluation model for the design quality of online shopping websites.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Electronic Commerce Research and Applications 12, 2 (2013), 124–135.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [9] Yingshan Chang, Mridu Baldevraj Narang, Hisami Suzuki, Guihong Cao, Jianfeng Gao, and Yonatan Bisk.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' WebQA: Multihop and Multimodal QA.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' ArXiv abs/2109.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='00590 (2021).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [10] Gongwei Chen, Xinhang Song, Haitao Zeng, and Shuqiang Jiang.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Scene Recognition With Prototype-Agnostic Scene Layout.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' IEEE Transactions on Image Processing 29 (2020), 5877–5888.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [11] Kaize Ding, Jianling Wang, Jundong Li, Dingcheng Li, and Huan Liu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Be More with Less: Hypergraph Attention Networks for Inductive Text Classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' ArXiv abs/2011.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='00387 (2020).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [12] Justin Gilmer, Samuel S Schoenholz, Patrick F Riley, Oriol Vinyals, and George E Dahl.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2017.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Neural message passing for quantum chemistry.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In International conference on machine learning.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' PMLR, 1263–1272.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [13] Jiafeng Guo, Yinqiong Cai, Yixing Fan, Fei Sun, Ruqing Zhang, and Xueqi Cheng.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Semantic models for the first-stage retrieval: A comprehensive review.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' arXiv preprint arXiv:2103.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='04831 (2021).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [14] Divya Gupta.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' An empirical study of predictive model for website quality analytics using dataset of different domains of websites.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Recent Advances in Computer Science and Communications (Formerly: Recent Patents on Computer Science) 14, 4 (2021), 995–1007.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [15] Will Hamilton, Zhitao Ying, and Jure Leskovec.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2017.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Inductive representation learning on large graphs.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Advances in neural information processing systems 30 (2017).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [16] Layla Hasan and Emad Abuelrub.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2011.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Assessing the quality of web sites.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Applied computing and informatics 9, 1 (2011), 11–29.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [17] Daniel Hasan Dalip, Marcos André Gonçalves, Marco Cristo, and Pável Calado.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2009.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Automatic quality assessment of content created collaboratively by web communities: a case study of wikipedia.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In Proceedings of the 9th ACM/IEEE-CS joint conference on Digital libraries.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 295–304.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [18] Lianzhe Huang, Dehong Ma, Sujian Li, Xiaodong Zhang, and Houfeng Wang.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2019.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Text level graph neural network for text classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' arXiv preprint arXiv:1910.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='02356 (2019).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [19] Lianzhe Huang, Dehong Ma, Sujian Li, Xiaodong Zhang, and Houfeng Wang.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2019.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Text Level Graph Neural Network for Text Classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In EMNLP.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [20] Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, and Furu Wei.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2022.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' ArXiv abs/2204.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='08387 (2022).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [21] Thomas Kipf and Max Welling.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2017.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Semi-Supervised Classification with Graph Convolutional Networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' ArXiv abs/1609.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='02907 (2017).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [22] Yann LeCun, Yoshua Bengio, and Geoffrey Hinton.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2015.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Deep learning.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' nature 521, 7553 (2015), 436–444.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [23] Yiding Liu, Weixue Lu, Suqi Cheng, Daiting Shi, Shuaiqiang Wang, Zhicong Cheng, and Dawei Yin.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Pre-trained language model for web-scale retrieval in baidu search.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 3365–3375.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [24] Vassilis Moustakis, Charalambos Litos, Andreas Dalivigas, Loukas Tsironis, et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2004.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Website Quality Assessment Criteria.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='. In ICIQ.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 59–73.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [25] Vahid Rafe and Maryam Monfaredzadeh.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2012.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' A qualitative framework to assess hospital/medical websites.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Journal of medical systems 36, 5 (2012), 2927–2939.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [26] Soumya Sanyal, Ivan V.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Anishchenko, Anirudh Dagar, David Baker, and Partha Pratim Talukdar.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' ProteinGCN: Protein model quality assessment using Graph Convolutional Networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' bioRxiv (2020).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [27] Artur M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Schweidtmann, Jan G.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Rittig, Andrea König, Martin Grohe, Alexander Mitsos, and Manuel Dahmen.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Graph Neural Networks for Prediction of Fuel Ignition Quality.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Energy & Fuels (2020).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [28] Yelong Shen, Xiaodong He, Jianfeng Gao, Li Deng, and Grégoire Mesnil.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2014.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Learning semantic representations using convolutional neural networks for web search.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In Proceedings of the 23rd international conference on world wide web.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 373–374.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [29] Kai Sheng Tai, Richard Socher, and Christopher D Manning.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2015.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Improved semantic representations from tree-structured long short-term memory networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' arXiv preprint arXiv:1503.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='00075 (2015).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [30] Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2017.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Graph attention networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' arXiv preprint arXiv:1710.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='10903 (2017).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [31] Te-Lin Wu, Cheng Li, Mingyang Zhang, Tao Chen, Spurthi Amba Hombaiah, and Michael Bendersky.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' LAMPRET: Layout-Aware Multimodal PreTraining for Document Understanding.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' arXiv preprint arXiv:2104.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='08405 (2021).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [32] Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and S Yu Philip.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' A comprehensive survey on graph neural networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' IEEE transactions on neural networks and learning systems 32, 1 (2020), 4–24.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [33] Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2018.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' How powerful are graph neural networks?' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' arXiv preprint arXiv:1810.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='00826 (2018).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [34] Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, and Ming Zhou.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' LayoutLM: Pre-training of Text and Layout for Document Image Understanding.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (2020).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [35] Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei A.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' F.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Florêncio, Cha Zhang, Wanxiang Che, Min Zhang, and Lidong Zhou.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' LayoutLMv2: Multi-modal Pre-training for Visually-rich Document Understanding.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' ArXiv abs/2012.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='14740 (2021).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [36] Zhiju Yang, Weiping Pei, Mon-Chu Chen, and Chuan Yue.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2022.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' WTAGRAPH: Web Tracking and Advertising Detection using Graph Neural Networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2022 IEEE Symposium on Security and Privacy (SP) (2022), 1540–1557.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [37] Liang Yao, Chengsheng Mao, and Yuan Luo.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2019.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Graph convolutional net- works for text classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In Proceedings of the AAAI conference on artificial intelligence, Vol.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 33.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 7370–7377.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [38] Dawei Yin, Yuening Hu, Jiliang Tang, Tim Daly, Mianwei Zhou, Hua Ouyang, Jianhui Chen, Changsung Kang, Hongbo Deng, Chikashi Nobata, et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2016.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Ranking relevance in yahoo search.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 323–332.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [39] Chengxuan Ying, Tianle Cai, Shengjie Luo, Shuxin Zheng, Guolin Ke, Di He, Yanming Shen, and Tie-Yan Liu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Do Transformers Really Perform Bad for Graph Representation?' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='. In NeurIPS.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [40] Junbi Zhang, Xu Ma, Shengen Zhang, Xianqiang Zheng, Rui Chen, Yihua Pan, Lisong Dong, Yayi Wei, and Gonzalo R.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Arce.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Lithography layout classifica- tion based on graph convolution network.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In Advanced Lithography.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [41] Yufeng Zhang, Xueli Yu, Zeyu Cui, Shu Wu, Zhongzhen Wen, and Liang Wang.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Every document owns its structure: Inductive text classification via graph neural networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' arXiv preprint arXiv:2004.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content='13826 (2020).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' [42] Lixin Zou, Shengqiang Zhang, Hengyi Cai, Dehong Ma, Suqi Cheng, Shuaiqiang Wang, Daiting Shi, Zhicong Cheng, and Dawei Yin.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' Pre-trained language model based ranking in Baidu search.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'} +page_content=' 4014–4022.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/89FLT4oBgHgl3EQfti_6/content/2301.12152v1.pdf'}