---
language:
- en
license: other
task_categories:
- text-retrieval
tags:
- information-retrieval
- msmarco
- tf-idf
size_categories:
- 1K<n<10K
configs:
- config_name: corpus
  data_files: corpus.jsonl
- config_name: queries
  data_files: queries.jsonl
- config_name: top100
  data_files: top100.jsonl
---

# MS MARCO Document Mini

Public subset of the MS MARCO document-ranking dataset. It contains
30 queries, 2993 documents, and 3000 query-document candidate rows.

## Files

- `corpus.jsonl`: `doc_id`, `url`, `title`, and `body`.
- `queries.jsonl`: `query_id` and the reformulated declarative text in `query`.
- `top100.jsonl`: candidate documents with their original Indri rank and score.
- Equivalent TSV files are included for simple use with pandas or Gensim.

The top-100 rows are retrieval candidates produced by `IndriQueryLikelihood`;
they are not binary relevance judgments.

## Source and license

This is a derived subset of the [MS MARCO document ranking
dataset](https://microsoft.github.io/msmarco/). Users must comply with the
original MS MARCO dataset terms and citation requirements. No ownership over the
source texts is claimed.