← home
Github: datasets/wikir.py

ir_datasets: WikIR

Index
  1. wikir
  2. wikir/en1k
  3. wikir/en1k/test
  4. wikir/en1k/training
  5. wikir/en1k/validation
  6. wikir/en59k
  7. wikir/en59k/test
  8. wikir/en59k/training
  9. wikir/en59k/validation
  10. wikir/en78k
  11. wikir/en78k/test
  12. wikir/en78k/training
  13. wikir/en78k/validation
  14. wikir/ens78k
  15. wikir/ens78k/test
  16. wikir/ens78k/training
  17. wikir/ens78k/validation
  18. wikir/es13k
  19. wikir/es13k/test
  20. wikir/es13k/training
  21. wikir/es13k/validation
  22. wikir/fr14k
  23. wikir/fr14k/test
  24. wikir/fr14k/training
  25. wikir/fr14k/validation
  26. wikir/it16k
  27. wikir/it16k/test
  28. wikir/it16k/training
  29. wikir/it16k/validation

"wikir"

A suite of IR benchmarks in multiple languages built from Wikipeida.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }

"wikir/en1k"

A small version of WikIR for English.

docs
370K docs

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k')
# Index wikir/en1k
indexer = pt.IterDictIndexer('./indices/wikir_en1k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 369721,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  }
}

"wikir/en1k/test"

Test set of wikir/en1k. Scoreddocs are the provided BM25 run.

queries
100 queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/test")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/test queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/test')
index_ref = pt.IndexRef.of('./indices/wikir_en1k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
370K docs

Inherits docs from wikir/en1k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/test")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/test docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/test')
# Index wikir/en1k
indexer = pt.IterDictIndexer('./indices/wikir_en1k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
4.4K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence4.3K97.7%
2Query is the article title100 2.3%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/test")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/test qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/test')
index_ref = pt.IndexRef.of('./indices/wikir_en1k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
10K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/test")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/test scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/test')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 369721,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 100
  },
  "qrels": {
    "count": 4435,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 100,
          "1": 4335
        }
      }
    }
  },
  "scoreddocs": {
    "count": 10000
  }
}

"wikir/en1k/training"

Training set of wikir/en1k. Scoreddocs are the provided BM25 run.

queries
1.4K queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/training")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/training queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/training')
index_ref = pt.IndexRef.of('./indices/wikir_en1k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
370K docs

Inherits docs from wikir/en1k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/training")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/training docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/training')
# Index wikir/en1k
indexer = pt.IterDictIndexer('./indices/wikir_en1k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
48K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence46K97.0%
2Query is the article title1.4K3.0%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/training")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/training qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/training')
index_ref = pt.IndexRef.of('./indices/wikir_en1k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
144K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/training")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/training scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/training')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 369721,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 1444
  },
  "qrels": {
    "count": 47699,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 1444,
          "1": 46255
        }
      }
    }
  },
  "scoreddocs": {
    "count": 144400
  }
}

"wikir/en1k/validation"

Validation set of wikir/en1k. Scoreddocs are the provided BM25 run.

queries
100 queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/validation")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/validation queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/validation')
index_ref = pt.IndexRef.of('./indices/wikir_en1k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
370K docs

Inherits docs from wikir/en1k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/validation")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/validation docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/validation')
# Index wikir/en1k
indexer = pt.IterDictIndexer('./indices/wikir_en1k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
5.0K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence4.9K98.0%
2Query is the article title100 2.0%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/validation")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/validation qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/validation')
index_ref = pt.IndexRef.of('./indices/wikir_en1k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
10K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en1k/validation")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en1k/validation scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en1k/validation')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 369721,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 100
  },
  "qrels": {
    "count": 4979,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 100,
          "1": 4879
        }
      }
    }
  },
  "scoreddocs": {
    "count": 10000
  }
}

"wikir/en59k"

WikIR for English.

docs
2.5M docs

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k')
# Index wikir/en59k
indexer = pt.IterDictIndexer('./indices/wikir_en59k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2454785,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  }
}

"wikir/en59k/test"

Test set of wikir/en59k. Scoreddocs are the provided BM25 run.

queries
1.0K queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/test")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/test queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/test')
index_ref = pt.IndexRef.of('./indices/wikir_en59k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
2.5M docs

Inherits docs from wikir/en59k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/test")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/test docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/test')
# Index wikir/en59k
indexer = pt.IterDictIndexer('./indices/wikir_en59k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
105K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence104K99.0%
2Query is the article title1.0K1.0%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/test")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/test qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/test')
index_ref = pt.IndexRef.of('./indices/wikir_en59k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
100K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/test")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/test scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/test')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2454785,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 1000
  },
  "qrels": {
    "count": 104715,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 1000,
          "1": 103715
        }
      }
    }
  },
  "scoreddocs": {
    "count": 100000
  }
}

"wikir/en59k/training"

Training set of wikir/en59k. Scoreddocs are the provided BM25 run.

queries
57K queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/training")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/training queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/training')
index_ref = pt.IndexRef.of('./indices/wikir_en59k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
2.5M docs

Inherits docs from wikir/en59k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/training")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/training docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/training')
# Index wikir/en59k
indexer = pt.IterDictIndexer('./indices/wikir_en59k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
2.4M qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence2.4M97.7%
2Query is the article title57K2.3%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/training")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/training qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/training')
index_ref = pt.IndexRef.of('./indices/wikir_en59k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
5.7M scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/training")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/training scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/training')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2454785,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 57251
  },
  "qrels": {
    "count": 2443383,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 57251,
          "1": 2386132
        }
      }
    }
  },
  "scoreddocs": {
    "count": 5725100
  }
}

"wikir/en59k/validation"

Validation set of wikir/en59k. Scoreddocs are the provided BM25 run.

queries
1.0K queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/validation")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/validation queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/validation')
index_ref = pt.IndexRef.of('./indices/wikir_en59k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
2.5M docs

Inherits docs from wikir/en59k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/validation")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/validation docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/validation')
# Index wikir/en59k
indexer = pt.IterDictIndexer('./indices/wikir_en59k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
69K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence68K98.5%
2Query is the article title1.0K1.5%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/validation")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/validation qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/validation')
index_ref = pt.IndexRef.of('./indices/wikir_en59k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
100K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en59k/validation")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en59k/validation scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en59k/validation')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2454785,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 1000
  },
  "qrels": {
    "count": 68905,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 1000,
          "1": 67905
        }
      }
    }
  },
  "scoreddocs": {
    "count": 100000
  }
}

"wikir/en78k"

WikIR for English. This is one of the two versions used in Frej2020Wikir.

docs
2.5M docs

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k')
# Index wikir/en78k
indexer = pt.IterDictIndexer('./indices/wikir_en78k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2456637,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  }
}

"wikir/en78k/test"

Test set of wikir/en78k. Scoreddocs are the provided BM25 run.

queries
7.9K queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/test")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/test queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/test')
index_ref = pt.IndexRef.of('./indices/wikir_en78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
2.5M docs

Inherits docs from wikir/en78k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/test")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/test docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/test')
# Index wikir/en78k
indexer = pt.IterDictIndexer('./indices/wikir_en78k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
353K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence345K97.8%
2Query is the article title7.9K2.2%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/test")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/test qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/test')
index_ref = pt.IndexRef.of('./indices/wikir_en78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
786K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/test")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/test scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/test')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2456637,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 7862
  },
  "qrels": {
    "count": 353060,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 7862,
          "1": 345198
        }
      }
    }
  },
  "scoreddocs": {
    "count": 785600
  }
}

"wikir/en78k/training"

Training set of wikir/en78k. Scoreddocs are the provided BM25 run.

queries
63K queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/training")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/training queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/training')
index_ref = pt.IndexRef.of('./indices/wikir_en78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
2.5M docs

Inherits docs from wikir/en78k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/training")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/training docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/training')
# Index wikir/en78k
indexer = pt.IterDictIndexer('./indices/wikir_en78k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
2.4M qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence2.4M97.4%
2Query is the article title63K2.6%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/training")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/training qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/training')
index_ref = pt.IndexRef.of('./indices/wikir_en78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
6.3M scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/training")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/training scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/training')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2456637,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 62904
  },
  "qrels": {
    "count": 2435257,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 62904,
          "1": 2372353
        }
      }
    }
  },
  "scoreddocs": {
    "count": 6284800
  }
}

"wikir/en78k/validation"

Validation set of wikir/en78k. Scoreddocs are the provided BM25 run.

queries
7.9K queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/validation")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/validation queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/validation')
index_ref = pt.IndexRef.of('./indices/wikir_en78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
2.5M docs

Inherits docs from wikir/en78k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/validation")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/validation docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/validation')
# Index wikir/en78k
indexer = pt.IterDictIndexer('./indices/wikir_en78k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
272K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence264K97.1%
2Query is the article title7.9K2.9%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/validation")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/validation qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/validation')
index_ref = pt.IndexRef.of('./indices/wikir_en78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
786K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/en78k/validation")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/en78k/validation scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/en78k/validation')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2456637,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 7862
  },
  "qrels": {
    "count": 271874,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 7862,
          "1": 264012
        }
      }
    }
  },
  "scoreddocs": {
    "count": 785700
  }
}

"wikir/ens78k"

WikIR for English, using the first sentences of articles as queries. This is one of the two versions used in Frej2020Wikir.

docs
2.5M docs

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k')
# Index wikir/ens78k
indexer = pt.IterDictIndexer('./indices/wikir_ens78k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2456637,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  }
}

"wikir/ens78k/test"

Test set of wikir/ens78k. Scoreddocs are the provided BM25 run.

queries
7.9K queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/test")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/test queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/test')
index_ref = pt.IndexRef.of('./indices/wikir_ens78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
2.5M docs

Inherits docs from wikir/ens78k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/test")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/test docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/test')
# Index wikir/ens78k
indexer = pt.IterDictIndexer('./indices/wikir_ens78k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
353K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence345K97.8%
2Query is the article title7.9K2.2%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/test")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/test qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/test')
index_ref = pt.IndexRef.of('./indices/wikir_ens78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
786K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/test")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/test scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/test')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2456637,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 7862
  },
  "qrels": {
    "count": 353060,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 7862,
          "1": 345198
        }
      }
    }
  },
  "scoreddocs": {
    "count": 786100
  }
}

"wikir/ens78k/training"

Training set of wikir/ens78k. Scoreddocs are the provided BM25 run.

queries
63K queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/training")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/training queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/training')
index_ref = pt.IndexRef.of('./indices/wikir_ens78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
2.5M docs

Inherits docs from wikir/ens78k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/training")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/training docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/training')
# Index wikir/ens78k
indexer = pt.IterDictIndexer('./indices/wikir_ens78k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
2.4M qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence2.4M97.4%
2Query is the article title63K2.6%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/training")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/training qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/training')
index_ref = pt.IndexRef.of('./indices/wikir_ens78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
6.3M scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/training")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/training scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/training')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2456637,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 62904
  },
  "qrels": {
    "count": 2435257,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 62904,
          "1": 2372353
        }
      }
    }
  },
  "scoreddocs": {
    "count": 6289800
  }
}

"wikir/ens78k/validation"

Validation set of wikir/ens78k. Scoreddocs are the provided BM25 run.

queries
7.9K queries

Language: en

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/validation")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/validation queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/validation')
index_ref = pt.IndexRef.of('./indices/wikir_ens78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pipeline(dataset.get_topics())

You can find more details about PyTerrier retrieval here.

docs
2.5M docs

Inherits docs from wikir/ens78k

Language: en

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/validation")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/validation docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/validation')
# Index wikir/ens78k
indexer = pt.IterDictIndexer('./indices/wikir_ens78k')
index_ref = indexer.index(dataset.get_corpus_iter(), fields=['text'])

You can find more details about PyTerrier indexing here.

qrels
272K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence264K97.1%
2Query is the article title7.9K2.9%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/validation")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/validation qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
from pyterrier.measures import *
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/validation')
index_ref = pt.IndexRef.of('./indices/wikir_ens78k') # assumes you have already built an index
pipeline = pt.BatchRetrieve(index_ref, wmodel='BM25')
# (optionally other pipeline components)
pt.Experiment(
    [pipeline],
    dataset.get_topics(),
    dataset.get_qrels(),
    [MAP, nDCG@20]
)

You can find more details about PyTerrier experiments here.

scoreddocs
786K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/ens78k/validation")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/ens78k/validation scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier 
import pyterrier as pt
pt.init()
dataset = pt.get_dataset('irds:wikir/ens78k/validation')
dataset.get_results()

You can find more details about PyTerrier dataset API here.

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 2456637,
    "fields": {
      "doc_id": {
        "max_len": 7,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 7862
  },
  "qrels": {
    "count": 271874,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 7862,
          "1": 264012
        }
      }
    }
  },
  "scoreddocs": {
    "count": 786100
  }
}

"wikir/es13k"

WikIR for Spanish.

docs
646K docs

Language: es

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 645901,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  }
}

"wikir/es13k/test"

Test set of wikir/es13k. Scoreddocs are the provided BM25 run.

queries
1.3K queries

Language: es

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/test")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/test queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

docs
646K docs

Inherits docs from wikir/es13k

Language: es

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/test")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/test docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

qrels
71K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence70K98.2%
2Query is the article title1.3K1.8%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/test")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/test qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

scoreddocs
130K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/test")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/test scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 645901,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 1300
  },
  "qrels": {
    "count": 71339,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 1300,
          "1": 70039
        }
      }
    }
  },
  "scoreddocs": {
    "count": 130000
  }
}

"wikir/es13k/training"

Training set of wikir/es13k. Scoreddocs are the provided BM25 run.

queries
11K queries

Language: es

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/training")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/training queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

docs
646K docs

Inherits docs from wikir/es13k

Language: es

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/training")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/training docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

qrels
477K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence466K97.7%
2Query is the article title11K2.3%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/training")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/training qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

scoreddocs
1.1M scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/training")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/training scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 645901,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 11202
  },
  "qrels": {
    "count": 477212,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 11202,
          "1": 466010
        }
      }
    }
  },
  "scoreddocs": {
    "count": 1120200
  }
}

"wikir/es13k/validation"

Validation set of wikir/es13k. Scoreddocs are the provided BM25 run.

queries
1.3K queries

Language: es

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/validation")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/validation queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

docs
646K docs

Inherits docs from wikir/es13k

Language: es

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/validation")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/validation docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

qrels
59K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence57K97.8%
2Query is the article title1.3K2.2%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/validation")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/validation qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

scoreddocs
130K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/es13k/validation")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/es13k/validation scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 645901,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 1300
  },
  "qrels": {
    "count": 58757,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 1300,
          "1": 57457
        }
      }
    }
  },
  "scoreddocs": {
    "count": 130000
  }
}

"wikir/fr14k"

WikIR for French.

docs
737K docs

Language: fr

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 736616,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  }
}

"wikir/fr14k/test"

Test set of wikir/fr14k. Scoreddocs are the provided BM25 run.

queries
1.4K queries

Language: fr

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/test")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/test queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

docs
737K docs

Inherits docs from wikir/fr14k

Language: fr

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/test")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/test docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

qrels
56K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence54K97.5%
2Query is the article title1.4K2.5%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/test")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/test qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

scoreddocs
140K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/test")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/test scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 736616,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 1400
  },
  "qrels": {
    "count": 55647,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 1400,
          "1": 54247
        }
      }
    }
  },
  "scoreddocs": {
    "count": 140000
  }
}

"wikir/fr14k/training"

Training set of wikir/fr14k. Scoreddocs are the provided BM25 run.

queries
11K queries

Language: fr

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/training")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/training queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

docs
737K docs

Inherits docs from wikir/fr14k

Language: fr

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/training")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/training docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

qrels
609K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence598K98.1%
2Query is the article title11K1.9%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/training")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/training qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

scoreddocs
1.1M scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/training")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/training scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 736616,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 11341
  },
  "qrels": {
    "count": 609240,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 11341,
          "1": 597899
        }
      }
    }
  },
  "scoreddocs": {
    "count": 1134100
  }
}

"wikir/fr14k/validation"

Validation set of wikir/fr14k. Scoreddocs are the provided BM25 run.

queries
1.4K queries

Language: fr

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/validation")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/validation queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

docs
737K docs

Inherits docs from wikir/fr14k

Language: fr

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/validation")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/validation docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

qrels
81K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence80K98.3%
2Query is the article title1.4K1.7%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/validation")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/validation qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

scoreddocs
140K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/fr14k/validation")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/fr14k/validation scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 736616,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 1400
  },
  "qrels": {
    "count": 81255,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 1400,
          "1": 79855
        }
      }
    }
  },
  "scoreddocs": {
    "count": 140000
  }
}

"wikir/it16k"

WikIR for Italian.

docs
503K docs

Language: it

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 503012,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  }
}

"wikir/it16k/test"

Test set of wikir/it16k. Scoreddocs are the provided BM25 run.

queries
1.6K queries

Language: it

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/test")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/test queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

docs
503K docs

Inherits docs from wikir/it16k

Language: it

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/test")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/test docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

qrels
49K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence48K96.8%
2Query is the article title1.6K3.2%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/test")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/test qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

scoreddocs
160K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/test")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/test scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 503012,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 1600
  },
  "qrels": {
    "count": 49338,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 1600,
          "1": 47738
        }
      }
    }
  },
  "scoreddocs": {
    "count": 160000
  }
}

"wikir/it16k/training"

Training set of wikir/it16k. Scoreddocs are the provided BM25 run.

queries
13K queries

Language: it

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/training")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/training queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

docs
503K docs

Inherits docs from wikir/it16k

Language: it

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/training")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/training docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

qrels
382K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence369K96.5%
2Query is the article title13K3.5%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/training")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/training qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

scoreddocs
1.3M scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/training")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/training scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 503012,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 13418
  },
  "qrels": {
    "count": 381920,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 13418,
          "1": 368502
        }
      }
    }
  },
  "scoreddocs": {
    "count": 1341800
  }
}

"wikir/it16k/validation"

Validation set of wikir/it16k. Scoreddocs are the provided BM25 run.

queries
1.6K queries

Language: it

Query type:
GenericQuery: (namedtuple)
  1. query_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/validation")
for query in dataset.queries_iter():
    query # namedtuple<query_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/validation queries
[query_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

docs
503K docs

Inherits docs from wikir/it16k

Language: it

Document type:
GenericDoc: (namedtuple)
  1. doc_id: str
  2. text: str

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/validation")
for doc in dataset.docs_iter():
    doc # namedtuple<doc_id, text>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/validation docs
[doc_id]    [text]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

qrels
45K qrels
Query relevance judgment type:
TrecQrel: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. relevance: int
  4. iteration: str

Relevance levels

Rel.DefinitionCount%
0Otherwise0 0.0%
1There is a link to the article with the query as its title in the first sentence43K96.4%
2Query is the article title1.6K3.6%

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/validation")
for qrel in dataset.qrels_iter():
    qrel # namedtuple<query_id, doc_id, relevance, iteration>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/validation qrels --format tsv
[query_id]    [doc_id]    [relevance]    [iteration]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

scoreddocs
160K scoreddocs
Scored Document type:
GenericScoredDoc: (namedtuple)
  1. query_id: str
  2. doc_id: str
  3. score: float

Examples:

Python API 
import ir_datasets
dataset = ir_datasets.load("wikir/it16k/validation")
for scoreddoc in dataset.scoreddocs_iter():
    scoreddoc # namedtuple<query_id, doc_id, score>

You can find more details about the Python API here.

CLI 
ir_datasets export wikir/it16k/validation scoreddocs --format tsv
[query_id]    [doc_id]    [score]
...

You can find more details about the CLI here.

PyTerrier

No example available for PyTerrier

Citation

ir_datasets.bib:

\cite{Frej2020Wikir,Frej2020MlWikir}

Bibtex:

@inproceedings{Frej2020Wikir, title={WIKIR: A Python toolkit for building a large-scale Wikipedia-based English Information Retrieval Dataset}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={LREC}, year={2020} } @inproceedings{Frej2020MlWikir, title={MLWIKIR: A Python Toolkit for Building Large-scale Wikipedia-based Information Retrieval Datasets in Chinese, English, French, Italian, Japanese, Spanish and More}, author={Jibril Frej and Didier Schwab and Jean-Pierre Chevallet}, booktitle={CIRCLE}, year={2020} }
Metadata 
{
  "docs": {
    "count": 503012,
    "fields": {
      "doc_id": {
        "max_len": 6,
        "common_prefix": ""
      }
    }
  },
  "queries": {
    "count": 1600
  },
  "qrels": {
    "count": 45003,
    "fields": {
      "relevance": {
        "counts_by_value": {
          "2": 1600,
          "1": 43403
        }
      }
    }
  },
  "scoreddocs": {
    "count": 160000
  }
}