💾 ir-datasets-longeval

Extension for accessing the LongEval datasets via ir_datasets.

Installation

Install the package from PyPI:

pip install ir-datasets-longeval

Usage

[!TIP] LongEval 2026: The new longeval-sci-2026 test collection is now available and we included extra tags for the shared task: longeval-sci-2026/clef-2026/sci and longeval-sci-2026/clef-2026/rag.

The ir_datasets_longeval extension provides a load method that returns a LongEval ir_dataset that allows to load official versions of the LongEval datasets as well as modified versions that you have on your local file system:

from ir_datasets_longeval import load

# load an official version of the LongEval dataset.
dataset = load("longeval-sci-2026/snapshot-3")

# load a local copy of a LongEval dataset.
# E.g., so that you can easily run your approach on modified data.
dataset = load("<PATH-TO-A-DIRECTORY-ON-YOUR-MACHINE>")

# From now on, you can use dataset as any ir_dataset

LongEval datasets have a set of temporal specifics that you can use:

# At what time does/did a dataset take place?
dataset.get_timestamp()

# Each dataset can have a list of zero or more past datasets/interactions.
# You can incorporate them in your retrieval system:
for past_dataset in dataset.get_prior_datasets():
    # `past_dataset` is an LongEval `ir_dataset` with the same functionality as the `dataset`
    past_dataset.get_timestamp()

If you want to use the CLI, just use the ir_datasets_longeval instead of ir_datasets. All CLI commands will work as usual, e.g., to list the officially available datasets:

ir_datasets_longeval list

Datasets

LongEval 2026

Details

The fourth LongEval Lab in 2026 introduced a new LongEval-Sci test collection. It contains three snapshots that each span three months. snapshot-1 from March to May, snapshot-2 from June to August, snapshot-3 from September to November, all in 2025. Additionally, training queries and qrels are provided for snapshot-1. Each snapshot contains different qrels sets: raw qrels mark all clicked documents as relevant and dctr qrels use the Document Click Through Rate (DCTR) as pseudo relevance label. Additionally, a set of RAG questions is available for the most recent snapshot snapshot-3.

Meta Tags:

longeval-sci-2026/*
longeval-sci-2026/clef-2026/sci
longeval-sci-2026/clef-2026/sci/raw
longeval-sci-2026/clef-2026/sci/dctr

LongEval Sci

Details

The third LongEval Lab introduced the first LongEval-Sci test collection. It contains the two snapshots 2024-11 and 2025-01 and additional training queries and qrels for 2024-11.

Meta Tags:

longeval-sci/*
longeval-sci/clef-2025-test

Citation:

@inproceedings{DBLP:conf/ecir/AlkhalifaBDEAFSGGILMMMPPSZ24,
  author       = {Rabab Alkhalifa and
                  Hsuvas Borkakoty and
                  Romain Deveaud and
                  Alaa El{-}Ebshihy and
                  Luis Espinosa Anke and
                  Tobias Fink and
                  Gabriela Gonz{\'{a}}lez S{\'{a}}ez and
                  Petra Galusc{\'{a}}kov{\'{a}} and
                  Lorraine Goeuriot and
                  David Iommi and
                  Maria Liakata and
                  Harish Tayyar Madabushi and
                  Pablo Medina{-}Alias and
                  Philippe Mulhem and
                  Florina Piroi and
                  Martin Popel and
                  Christophe Servan and
                  Arkaitz Zubiaga},
  editor       = {Nazli Goharian and
                  Nicola Tonellotto and
                  Yulan He and
                  Aldo Lipani and
                  Graham McDonald and
                  Craig Macdonald and
                  Iadh Ounis},
  title        = {LongEval: Longitudinal Evaluation of Model Performance at {CLEF} 2024},
  booktitle    = {Advances in Information Retrieval - 46th European Conference on Information
                  Retrieval, {ECIR} 2024, Glasgow, UK, March 24-28, 2024, Proceedings,
                  Part {VI}},
  series       = {Lecture Notes in Computer Science},
  volume       = {14613},
  pages        = {60--66},
  publisher    = {Springer},
  year         = {2024},
  url          = {https://doi.org/10.1007/978-3-031-56072-9\_8},
  doi          = {10.1007/978-3-031-56072-9\_8},
  timestamp    = {Mon, 15 Apr 2024 08:25:15 +0200},
  biburl       = {https://dblp.org/rec/conf/ecir/AlkhalifaBDEAFSGGILMMMPPSZ24.bib},
  bibsource    = {dblp computer science bibliography, https://dblp.org}
}

LongEval Web

Details

The third LongEval Lab continued the LongEval test collection and made many more snapshots available. It is only available in French and contains the monthly snapshots 2022-06, 2022-07, 2022-08, 2022-09, 2022-10, 2022-11, 2022-12, 2023-01, 2023-02, 2023-03, 2023-04, 2023-05, 2023-06, 2023-07, 2023-08.

Meta Tags:

longeval-web/*
longeval-web/clef-2025-test

Citation:

@inproceedings{DBLP:conf/ecir/AlkhalifaBDEAFSGGILMMMPPSZ24,
  author       = {Rabab Alkhalifa and
                  Hsuvas Borkakoty and
                  Romain Deveaud and
                  Alaa El{-}Ebshihy and
                  Luis Espinosa Anke and
                  Tobias Fink and
                  Gabriela Gonz{\'{a}}lez S{\'{a}}ez and
                  Petra Galusc{\'{a}}kov{\'{a}} and
                  Lorraine Goeuriot and
                  David Iommi and
                  Maria Liakata and
                  Harish Tayyar Madabushi and
                  Pablo Medina{-}Alias and
                  Philippe Mulhem and
                  Florina Piroi and
                  Martin Popel and
                  Christophe Servan and
                  Arkaitz Zubiaga},
  editor       = {Nazli Goharian and
                  Nicola Tonellotto and
                  Yulan He and
                  Aldo Lipani and
                  Graham McDonald and
                  Craig Macdonald and
                  Iadh Ounis},
  title        = {LongEval: Longitudinal Evaluation of Model Performance at {CLEF} 2024},
  booktitle    = {Advances in Information Retrieval - 46th European Conference on Information
                  Retrieval, {ECIR} 2024, Glasgow, UK, March 24-28, 2024, Proceedings,
                  Part {VI}},
  series       = {Lecture Notes in Computer Science},
  volume       = {14613},
  pages        = {60--66},
  publisher    = {Springer},
  year         = {2024},
  url          = {https://doi.org/10.1007/978-3-031-56072-9\_8},
  doi          = {10.1007/978-3-031-56072-9\_8},
  timestamp    = {Mon, 15 Apr 2024 08:25:15 +0200},
  biburl       = {https://dblp.org/rec/conf/ecir/AlkhalifaBDEAFSGGILMMMPPSZ24.bib},
  bibsource    = {dblp computer science bibliography, https://dblp.org}
}

LongEval 2023

The dataset is hosted at Lindat and a local copy is needed to be placed in the ir_dataset directory.

Details

The original LongEval test collection is available in French and English and comprises the three snapshots 2022-06 (WT), 2022-07 (ST), and 2022-08 (LT). The initial version assigned different IDs to documents and queries present in multiple snapshots. The original IDs are available in the non-unified tags.

Citation:

@inproceedings{DBLP:conf/sigir/GaluscakovaDSMG23,
  author       = {Petra Galusc{\'{a}}kov{\'{a}} and
                  Romain Deveaud and
                  Gabriela Gonz{\'{a}}lez S{\'{a}}ez and
                  Philippe Mulhem and
                  Lorraine Goeuriot and
                  Florina Piroi and
                  Martin Popel},
  editor       = {Hsin{-}Hsi Chen and
                  Wei{-}Jou (Edward) Duh and
                  Hen{-}Hsen Huang and
                  Makoto P. Kato and
                  Josiane Mothe and
                  Barbara Poblete},
  title        = {LongEval-Retrieval: French-English Dynamic Test Collection for Continuous
                  Web Search Evaluation},
  booktitle    = {Proceedings of the 46th International {ACM} {SIGIR} Conference on
                  Research and Development in Information Retrieval, {SIGIR} 2023, Taipei,
                  Taiwan, July 23-27, 2023},
  pages        = {3086--3094},
  publisher    = {{ACM}},
  year         = {2023},
  url          = {https://doi.org/10.1145/3539618.3591921},
  doi          = {10.1145/3539618.3591921},
  timestamp    = {Wed, 25 Feb 2026 08:28:08 +0100},
  biburl       = {https://dblp.org/rec/conf/sigir/GaluscakovaDSMG23.bib},
  bibsource    = {dblp computer science bibliography, https://dblp.org}
}

Citation

If you use this package, please cite the original ir_datasets paper and this extension:

@inproceedings{ir_datasets_longeval,
  author       = {J{\"{u}}ri Keller and Maik Fr{\"{o}}be and Gijs Hendriksen and Daria Alexander and Martin Potthast and Philipp Schaer},
  title        = {Simplified Longitudinal Retrieval Experiments: A Case Study on Query Expansion and Document Boosting},
  booktitle    = {Experimental {IR} Meets Multilinguality, Multimodality, and Interaction - 16th International Conference of the {CLEF} Association, {CLEF} 2024, Madrid, Spain, September 9-12, 2025, Proceedings, Part {I}},
  series       = {Lecture Notes in Computer Science},
  publisher    = {Springer},
  year         = {2025}
}

Development

To build this package and contribute to its development you need to install the build, setuptools, and wheel packages (pre-installed on most systems):

pip install build setuptools wheel

Create and activate a virtual environment:

python3.10 -m venv venv/
source venv/bin/activate

Dependencies

Install the package and test dependencies:

pip install -e .[tests]

Testing

Verify your changes against the test suite to verify.

ruff check .                   # Code format and LINT
mypy .                         # Static typing
bandit -c pyproject.toml -r .  # Security
pytest .                       # Unit tests

Please also add tests for your newly developed code.

Build wheels

Wheels for this package can be built with:

python -m build

Support

If you have any problems using this package, please file an issue. We're happy to help!

Fork Notice

This repository is a fork of ir-datasets-clueweb22, originally developed by Jan Heinrich Merker. All credit for the original work goes to him, and this fork retains the original MIT License. The changes made in this fork include an adaptation from the clueweb22 dataset to the LongEval datasets.

License

This repository is released under the MIT license.

Metadata

Release files for ir-datasets-longeval 0.0.15

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for ir-datasets-longeval 0.0.15
File	Size	Uploaded
ir_datasets_longeval-0.0.15.tar.gz	110.6 kB	May 10, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for ir-datasets-longeval 0.0.15
File	Interpreter	ABI	Platform	Reset
ir_datasets_longeval-0.0.15-py3-none-any.whl 27.8 kB May 10, 2026	Python 3	none	any	Details

Total release size: 138.4 kB

Release files / ir_datasets_longeval-0.0.15.tar.gz

Download URL	ir_datasets_longeval-0.0.15.tar.gz
Size	110.6 kB
Tags	Source
SHA-256 checksum How to use checksums	`fa3da0f6f4e4b694836e1127da66102820e4e73a474bd9600d2ff07200434031`
BLAKE2b-256 checksum How to use checksums	`34a4afcf3b8528fe48944abc0f00a70f9587737b144ad91105fd38e5d87bce4b`
Upload date	May 10, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/6.1.0 CPython/3.13.12`

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/clef-longeval/ir-datasets-longeval
Publishing commit: github.com/clef-longeval/ir-datasets-longeval/tree/6187814a982d43ca6fed76c73bf09f6345df1826

Workflow

Publishing configuration: github.com/clef-longeval/ir-datasets-longeval/blob/6187814a982d43ca6fed76c73bf09f6345df1826/.github/workflows/ci.yml
Publishing logs: github.com/clef-longeval/ir-datasets-longeval/actions/runs/25636024927/attempts/1

Artifact

Subject: ir_datasets_longeval-0.0.15.tar.gz
SHA-256 checksum: fa3da0f6f4e4b694836e1127da66102820e4e73a474bd9600d2ff07200434031 Verifying attestations

Release files / ir_datasets_longeval-0.0.15-py3-none-any.whl

Download URL	ir_datasets_longeval-0.0.15-py3-none-any.whl
Size	27.8 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`64105db6da3c2b37c896e5a78cddc720b26ee02d1362eb848a9c5ec172d07412`
BLAKE2b-256 checksum How to use checksums	`c6b5caef79aec809f104898687de1fb22cc980523d440f509672125bec2fa1c0`
Upload date	May 10, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/6.1.0 CPython/3.13.12`

Provenance

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/clef-longeval/ir-datasets-longeval
Publishing commit: github.com/clef-longeval/ir-datasets-longeval/tree/6187814a982d43ca6fed76c73bf09f6345df1826

Workflow

Publishing configuration: github.com/clef-longeval/ir-datasets-longeval/blob/6187814a982d43ca6fed76c73bf09f6345df1826/.github/workflows/ci.yml
Publishing logs: github.com/clef-longeval/ir-datasets-longeval/actions/runs/25636024927/attempts/1

Artifact

Subject: ir_datasets_longeval-0.0.15-py3-none-any.whl
SHA-256 checksum: 64105db6da3c2b37c896e5a78cddc720b26ee02d1362eb848a9c5ec172d07412 Verifying attestations