Skip to content

Installation

Install ir_axioms from PyPI

The ir_axioms framework requires Python 3.11 or later.

For a basic, minimal setup, just install the ir_axioms package from PyPI, and a spaCy model for tokenization:

pip install ir_axioms
python -m spacy download en_core_web_sm

If you do not need any "fancy" extras, just continue to the quickstart.

Optional Dependencies

Some features of ir_axioms require additional dependencies that you can install as "extras" of the PyPI package like so:

pip install ir_axioms[pyterrier,sbert]

All available extras are listed here for quick reference:

Extra Description
pyserini Use index statistics, document contents, and tokenization from Pyserini/Anserini indices.
pyterrier Use index statistics, document contents, and tokenization from PyTerrier/Terrier indices. Apply KwikSort re-ranking to PyTerrier pipelines. Run axiomatic analyses.
sbert Use Sentence Transformers for sentence similarity (enabled automatically if installed).
model2vec Use Model2Vec static embeddings for sentence similarity (enabled automatically if installed).
blingfire Use BlingFire for fast tokenization (enabled automatically if installed).
fasttext Use fastText word vectors for term similarity (enabled automatically if installed).
keybert Use KeyBERT for aspect extraction (must be enabled manually if installed).
autojudge Use axioms as TREC AutoJudge evaluators.
experiments Dependencies for the experiment notebooks.

Reproducing our ICTIR 2025 results

To reproduce the results from our ICTIR 2025 paper on axioms for RAG [Merker et al., 2025], please install these extras:

pip install ir_axioms[pyterrier,keybert,sbert,fasttext]

Command Line Interface

Installing the ir_axioms package also installs the ir_axioms CLI command (aliases: ir-axioms, axioms). Run its help command to list available sub-commands:

ir_axioms --help

Work in Progress

Right now, the CLI does not do a lot... see the roadmap.

Development Installation

To work on ir_axioms core code itself, please follow the contribution guide.

References

Jan Heinrich Merker, Maik Fröbe, Benno Stein, Martin Potthast, and Matthias Hagen. Axioms for Retrieval-Augmented Generation. In Proceedings of the 2025 International ACM SIGIR Conference on Innovative Concepts and Theories in Information Retrieval, ICTIR 2025, Padua, Italy, July 18, 2025, pages 67–77, 2025. ACM. [] [doi] 
@inproceedings{merker2025axioms,
  author       = {Jan Heinrich Merker and
                  Maik Fr{\"{o}}be and
                  Benno Stein and
                  Martin Potthast and
                  Matthias Hagen},
  title        = {Axioms for Retrieval-Augmented Generation},
  booktitle    = {Proceedings of the 2025 International {ACM} {SIGIR} Conference on
                  Innovative Concepts and Theories in Information Retrieval, {ICTIR}
                  2025, Padua, Italy, July 18, 2025},
  pages        = {67--77},
  publisher    = {{ACM}},
  year         = {2025},
  url          = {https://doi.org/10.1145/3731120.3744601},
  doi          = {10.1145/3731120.3744601}
}