Datasets:

TheFinAI
/

plutus-QA

---
dataset_info:
  features:
  - name: query
    dtype: string
  - name: answer
    dtype: string
  - name: text
    dtype: string
  - name: choices
    sequence: string
  - name: gold
    dtype: int64
  splits:
  - name: train
    num_bytes: 474769
    num_examples: 267
  - name: validation
    num_bytes: 74374
    num_examples: 48
  - name: test
    num_bytes: 387629
    num_examples: 225
  download_size: 358300
  dataset_size: 936772
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
  - split: test
    path: data/test-*
license: apache-2.0
task_categories:
- question-answering
language:
- gr
tags:
- finance
- qa
pretty_name: Plutus QA
size_categories:
- n<1K
---

----------------------------------------------------------------
# Dataset Card for Plutus QA

## Table of Contents

- [Table of Contents](#table-of-contents)
- [Dataset Description](#dataset-description)
  - [Dataset Summary](#dataset-summary)
  - [Supported Tasks](#supported-tasks)
  - [Languages](#languages)
- [Dataset Structure](#dataset-structure)
  - [Data Instances](#data-instances)
  - [Data Fields](#data-fields)
  - [Data Splits](#data-splits)
- [Dataset Creation](#dataset-creation)
  - [Curation Rationale](#curation-rationale)
  - [Source Data](#source-data)
  - [Annotations](#annotations)
  - [Personal and Sensitive Information](#personal-and-sensitive-information)
- [Considerations for Using the Data](#considerations-for-using-the-data)
  - [Social Impact of Dataset](#social-impact-of-dataset)
  - [Discussion of Biases](#discussion-of-biases)
  - [Other Known Limitations](#other-known-limitations)
- [Additional Information](#additional-information)
  - [Dataset Curators](#dataset-curators)
  - [Licensing Information](#licensing-information)
  - [Citation Information](#citation-information)
  - [Contributions](#contributions)

## Dataset Description

- **Homepage:** https://huggingface.co/collections/TheFinAI/plutus-benchmarking-greek-financial-llms-67bc718fb8d897c65f1e87db
- **Repository:** https://huggingface.co/datasets/TheFinAI/plutus-QA
- **Paper:** Plutus: Benchmarking Large Language Models in Low-Resource Greek Finance
- **Leaderboard:** https://huggingface.co/spaces/TheFinAI/Open-Greek-Financial-LLM-Leaderboard#/
- **Model:** https://huggingface.co/spaces/TheFinAI/plutus-8B-instruct

### Dataset Summary

Plutus QA is a question-answering dataset designed for financial applications in the Greek language. This resource contains a diverse set of queries, each paired with an answer, additional context text, a set of multiple-choice options, and a gold label index indicating the correct choice. By integrating textual context with a multiple-choice format, the dataset is aimed at benchmarking the capabilities of large language models in resolving financial questions in low-resource settings.

### Supported Tasks

- **Task:** Question Answering
- **Evaluation Metrics:** Accuracy

### Languages

- Greek

## Dataset Structure

### Data Instances

Each instance in the dataset consists of the following five fields:

- **query:** A question or prompt regarding financial matters.
- **answer:** The corresponding answer text paired with the query.
- **text:** Additional context or background information to support the query.
- **choices:** A sequence field containing multiple-choice answer options.
- **gold:** An integer field representing the index of the correct answer in the choices.

### Data Fields

- **query:** String – Represents the financial question or prompt.
- **answer:** String – Contains the answer aligned with the given query.
- **text:** String – Provides extra contextual information related to the query.
- **choices:** Sequence of strings – Lists all available answer options.
- **gold:** Int64 – Denotes the index of the correct answer from the choices.

### Data Splits

The dataset is organized into three splits:

- **Train:** 267 instances (474,769 bytes)
- **Validation:** 48 instances (74,374 bytes)
- **Test:** 225 instances (387,629 bytes)

## Dataset Creation

### Curation Rationale

The Plutus QA dataset was created to enable the evaluation of large language models on question-answering tasks within the financial domain for Greek language texts. Its design—including multiple-choice answers with additional context—aims to reflect complex financial decision-making and reasoning processes in a low-resource language environment.

### Source Data

#### Initial Data Collection and Normalization

The source data is derived from Greek university financial exams. Standardization and normalization procedures were applied to ensure consistency across queries, choices, and textual context.

#### Who are the Source Language Producers?

Greek university financial exams.

### Annotations

#### Annotation Process

Annotations were performed by domain experts proficient in both finance and linguistics. The process included verifying the correct answer for each query and marking the corresponding correct index among provided choices.

#### Who are the Annotators?

A team of financial analysts and linguists collaborated to ensure the annotations are accurate and reflective of real-world financial reasoning.

### Personal and Sensitive Information

This dataset is curated to exclude any personally identifiable information (PII) and only contains public financial text data necessary for question-answering tasks.

## Considerations for Using the Data

### Social Impact of Dataset

The Plutus QA dataset is instrumental in enhancing automated question-answering systems in the financial sector, particularly for Greek language applications. Improved QA systems can support better financial decision-making, increase the efficiency of financial services, and contribute to academic research in financial natural language processing.

### Discussion of Biases

- The domain-specific financial language may limit generalizability to non-financial question-answering tasks.
- Annotation subjectivity could introduce biases in determining the correct answer among multiple choices.
- The dataset's focus on Greek financial documents may not fully represent other financial or multilingual contexts.

### Other Known Limitations

- Pre-processing may be required to handle variations in question and answer formats.
- The dataset is specialized for the financial domain and may need adaptation for different QA tasks or domains.

## Additional Information

### Dataset Curators

- Xueqing Peng
- Triantafillos Papadopoulos
- Efstathia Soufleri
- Polydoros Giannouris
- Ruoyu Xiang
- Yan Wang
- Lingfei Qian
- Jimin Huang
- Qianqian Xie
- Sophia Ananiadou

The research is supported by NaCTeM, Archimedes RC, and The Fin AI.


### Licensing Information

- **License:** Apache License 2.0

### Citation Information

If you use this dataset in your research, please consider citing it as follows:

```bibtex
@misc{peng2025plutusbenchmarkinglargelanguage,
      title={Plutus: Benchmarking Large Language Models in Low-Resource Greek Finance}, 
      author={Xueqing Peng and Triantafillos Papadopoulos and Efstathia Soufleri and Polydoros Giannouris and Ruoyu Xiang and Yan Wang and Lingfei Qian and Jimin Huang and Qianqian Xie and Sophia Ananiadou},
      year={2025},
      eprint={2502.18772},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2502.18772}, 
}
```