trec-covid.terrier

Description

Terrier index for TREC-COVID

Usage

# Load the artifact
import pyterrier as pt
index = pt.Artifact.from_hf('pyterrier/trec-covid.terrier')
index.bm25()

Benchmarks

name	nDCG@10	R@1000
bm25	0.6295	0.4593
dph	0.631	0.4102

Reproduction

import pyterrier as pt
from tqdm import tqdm
import ir_datasets
dataset = ir_datasets.load('beir/trec-covid')
meta_docno_len = dataset.metadata()['docs']['fields']['doc_id']['max_len']
indexer = pt.IterDictIndexer("./trec-covid.terrier", meta={'docno': meta_docno_len, 'text': 4096})
docs = ({'docno': d.doc_id, 'text': d.default_text()} for d in tqdm(dataset.docs))
indexer.index(docs)

Metadata

{
  "type": "sparse_index",
  "format": "terrier",
  "package_hint": "python-terrier"
}