Datasets:

Hyukkyu
/

beir-trec-covid

license: apache-2.0
task_categories:
  - information-retrieval
  - text-retrieval
tags:
  - beir
  - trec-covid
  - information-retrieval
  - retrieval
  - search
dataset_info:
  - config_name: corpus
    features:
      - name: _id
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: string
      - name: metadata
        struct:
          - name: pubmed_id
            dtype: string
          - name: url
            dtype: string
    splits:
      - name: train
        num_bytes: 210286181
        num_examples: 171332
    download_size: 115532524
    dataset_size: 210286181
  - config_name: queries
    features:
      - name: _id
        dtype: string
      - name: text
        dtype: string
      - name: metadata
        struct:
          - name: narrative
            dtype: string
          - name: query
            dtype: string
    splits:
      - name: train
        num_bytes: 13952
        num_examples: 50
    download_size: 11365
    dataset_size: 13952
configs:
  - config_name: corpus
    data_files:
      - split: train
        path: corpus/train-*
  - config_name: queries
    data_files:
      - split: train
        path: queries/train-*

BEIR TREC-COVID Dataset (Migrated)

This is a migrated version of BeIR/trec-covid that is compatible with datasets library 4.0.0+.

Dataset Description

This dataset contains the trec-covid dataset from the BEIR benchmark, converted from the old script-based format to Parquet format.

Dataset Structure

Queries

Split 'queries': 50 examples
- Features: ['_id', 'text', 'metadata']
Total examples: 50

Corpus

Split 'corpus': 171,332 examples
- Features: ['_id', 'title', 'text', 'metadata']
Total examples: 171,332

Usage

from datasets import load_dataset

# Load queries (split: queries)
queries = load_dataset("Hyukkyu/beir-trec-covid", "queries", split="queries")

# Load corpus (split: corpus)
corpus = load_dataset("Hyukkyu/beir-trec-covid", "corpus", split="corpus")

Available Splits

Queries

queries: 50 examples

Corpus

corpus: 171,332 examples

Original Dataset

This dataset is migrated from: BeIR/trec-covid

Citation

If you use this dataset, please cite the original BEIR paper:

@article{thakur2021beir,
  title={BEIR: A Heterogeneous Benchmark for Zero-shot Evaluation of Information Retrieval Models},
  author={Thakur, Nandan and Reimers, Nils and Ruckle, Andreas and Srivastava, Abhishek and Gurevych, Iryna},
  journal={arXiv preprint arXiv:2104.08663},
  year={2021}
}