razaulhaq commited on
Commit
8b2b733
·
verified ·
1 Parent(s): 2c410ac

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +107 -0
README.md ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ task_categories:
4
+ - text-classification
5
+ - question-answering
6
+ - text-generation
7
+ tags:
8
+ - legal
9
+ size_categories:
10
+ - 10K<n<100K
11
+ ---
12
+ # EUR-Lex SDG-Annotated Dataset
13
+
14
+ ## Dataset Summary
15
+
16
+ The **EUR-Lex SDG-Annotated Dataset** enhances the original [**EUR-Lex**](https://huggingface.co/datasets/NLP-AUEB/eurlex/blob/main/README.md) dataset by mapping European Union legislation documents to the **United Nations Sustainable Development Goals (SDGs)**. Each document in this dataset has been annotated using a rule-based keyword matching approach, leveraging SDG-related terms and synonyms extracted from an Excel-based term matrix.
17
+
18
+ This dataset provides **researchers**, **policymakers**, and **data scientists** with a unique resource to analyze the alignment of EU legislation with global sustainability objectives, facilitating tasks such as **legal text classification**, **sustainability impact analysis**, and **policy research**.
19
+
20
+ ## Dataset Structure
21
+
22
+ **Annotation Method:** Rule-based keyword matching using SDG-related terms from [EU SDG Mapper](https://knowsdgs.jrc.ec.europa.eu/sdgmapper)
23
+
24
+ **Data Splits:**
25
+ - `train`
26
+ - `validation`
27
+ - `test`
28
+
29
+ ## Features
30
+
31
+ Each document in the dataset includes the following fields:
32
+
33
+ - **celex_id:** Unique identifier for the legal document.
34
+ - **title:** Title of the document.
35
+ - **text:** Full text of the document.
36
+ - **SDG 1** to **SDG 17:** Proportional coverage of each SDG, calculated based on keyword matches. The values represent the ratio of matched keywords for a specific SDG relative to total matches across all SDGs in the document.
37
+
38
+ ## Annotation Process
39
+
40
+ ### Keyword Extraction
41
+
42
+ SDG-related keywords were extracted from the [EU SDG Mapper](https://knowsdgs.jrc.ec.europa.eu/sdgmapper) and organized into a structured [Excel file](https://code.europa.eu/knowsdgs/sdgmapper_keywords/-/raw/main/keywords/term_matrix.xlsx). The file contains a list of terms and synonyms aligned with each SDG.
43
+
44
+ ### Rule-Based Matching
45
+
46
+ Each document from the EUR-Lex dataset was processed to identify occurrences of SDG-related keywords. The matching was **case-insensitive** and covered a wide range of **synonyms** for comprehensive coverage.
47
+
48
+ ### Coverage Calculation
49
+
50
+ For each document, the proportion of matched keywords for each SDG was calculated relative to the total number of matched keywords. This provides a **normalized representation** of SDG relevance for each document.
51
+
52
+ ## Annotation Process
53
+
54
+ - **Keyword Extraction**
55
+ SDG-related keywords were extracted from the [EU SDG Mapper](https://knowsdgs.jrc.ec.europa.eu/sdgmapper) and organized into a structured [Excel file](https://code.europa.eu/knowsdgs/sdgmapper_keywords/-/raw/main/keywords/term_matrix.xlsx). The file contains a list of terms and synonyms aligned with each SDG.
56
+
57
+ - **Rule-Based Matching**
58
+ Each document from the EUR-Lex dataset was processed to identify occurrences of SDG-related keywords. The matching was **case-insensitive** and covered a wide range of **synonyms** for comprehensive coverage.
59
+
60
+ - **Coverage Calculation**
61
+ For each document, the proportion of matched keywords for each SDG was calculated relative to the total number of matched keywords. This provides a **normalized representation** of SDG relevance for each document.
62
+
63
+ ## Usage Example
64
+
65
+ To load and explore the dataset using the Hugging Face `datasets` library:
66
+
67
+ ```python
68
+ from datasets import load_dataset
69
+
70
+ # Load the dataset
71
+ dataset = load_dataset('your-username/eurlex-sdg-annotated', split='train')
72
+
73
+ # View a sample document
74
+ print(dataset[0])
75
+ ```
76
+
77
+
78
+ ## Applications
79
+
80
+ - **Legal Document Classification:** Classify EU legislation based on their alignment with SDGs.
81
+
82
+ - **Policy Analysis:** Assess the focus of EU policies on sustainability goals.
83
+
84
+ - **Research on SDG Impact:** Study the integration of sustainability in legal frameworks.
85
+
86
+ - **Natural Language Processing:** Use the dataset for training models in text classification, keyword extraction, or topic modeling.
87
+
88
+ ## Citation
89
+
90
+ If you use this dataset in your research, please cite it as follows:
91
+
92
+ ```bibtex
93
+ @dataset{eurlex_sdg_annotated,
94
+ title = {EUR-Lex SDG-Annotated Dataset},
95
+ author = {Raza ul Haq},
96
+ year = {2025},
97
+ url = {https://huggingface.co/datasets/razaulhaq/eurlex-sdg-annotated}
98
+ }
99
+ ```
100
+
101
+ ## Acknowledgments
102
+
103
+ - **Original Dataset:** The base EUR-Lex dataset was created by the NLP Group at the Athens University of Economics and Business ([NLP-AUEB](https://huggingface.co/datasets/NLP-AUEB/eurlex/blob/main/README.md)).
104
+
105
+ - **SDG Keywords Source:** The SDG terminology and keywords were sourced from the [EU SDG Mapper](https://knowsdgs.jrc.ec.europa.eu/sdgmapper).
106
+
107
+ - **SDG Framework:** The SDG goals are derived from the **United Nations Sustainable Development Goals** framework.