README.md

metadata

license: apache-2.0
task_categories:
  - question-answering
  - multiple-choice
language:
  - en
tags:
  - biology
  - medical
formats:
  - csv
pretty_name: biomixQA
size_categories:
  - n<1K
configs:
  - config_name: mcq
    data_files: mcq_biomix.csv
    sep: ','
  - config_name: true_false
    data_files: true_false_biomix.csv
    sep: ','

BiomixQA Dataset

Overview

BiomixQA is a curated biomedical question-answering dataset comprising two distinct components:

1. Multiple Choice Questions (MCQ)
2. True/False Questions

This dataset has been utilized to validate the Knowledge Graph based Retrieval-Augmented Generation (KG-RAG) framework across different Large Language Models (LLMs). The diverse nature of questions in this dataset, spanning multiple choice and true/false formats, along with its coverage of various biomedical concepts, makes it particularly suitable for assessing the performance of KG-RAG framework.

Hence, this dataset is designed to support research and development in biomedical natural language processing, knowledge graph reasoning, and question-answering systems.

Dataset Description

• Repository: https://github.com/BaranziniLab/KG_RAG
• Paper: Biomedical knowledge graph-optimized prompt generation for large language models
• Point of Contact: Karthik Soman

Dataset Components

1. Multiple Choice Questions (MCQ)

• File: mcq_biomix.csv
• Size: 306 questions
• Format: Each question has five choices with a single correct answer

2. True/False Questions

• File: true_false_biomix.csv
• Size: 311 questions
• Format: Binary (True/False) questions

Potential Uses

1. Evaluating biomedical question-answering systems
2. Testing natural language processing models in the biomedical domain
3. Assessing retrieval capabilities of various RAG (Retrieval-Augmented Generation) frameworks
4. Supporting research in biomedical ontologies and knowledge graphs

Source Data

1. SPOKE: A large scale biomedical knowledge graph that consists of ~40 million biomedical concepts and ~140 million biologically meaningful relationships (Morris et al. 2023).
2. DisGeNET: Consolidates data about genes and genetic variants linked to human diseases from curated repositories, the GWAS catalog, animal models, and scientific literature (Piñero et al. 2016).
3. MONDO: Provides information about the ontological classification of Disease entities in the Open Biomedical Ontologies (OBO) format (Vasilevsky et al. 2022).
4. SemMedDB: Contains semantic predications extracted from PubMed citations (Kilicoglu et al. 2012).
5. Monarch Initiative: A platform for disease-gene association data (Mungall et al. 2017).
6. ROBOKOP: A knowledge graph-based system for biomedical data integration and analysis (Bizon et al. 2019).

Citation

If you use this dataset in your research, please cite the following paper:

@article{soman2023biomedical,
  title={Biomedical knowledge graph-enhanced prompt generation for large language models},
  author={Soman, Karthik and Rose, Peter W and Morris, John H and Akbas, Rabia E and Smith, Brett and Peetoom, Braian and Villouta-Reyes, Catalina and Cerono, Gabriel and Shi, Yongmei and Rizk-Jackson, Angela and others},
  journal={arXiv preprint arXiv:2311.17330},
  year={2023}
}