← smac.pub home

PyTerrier: Declarative Experimentation in Python from BM25 to Dense Retrieval

bibtex resource conference paper to appear

Authors: Craig Macdonald, Nicola Tonellotto, Sean MacAvaney, Iadh Ounis

Appearing in: CIKM 2021

Abstract:

PyTerrier is a Python-based retrieval framework for expressing simple and complex information retrieval (IR) pipelines in a declarative manner. While making use of the long-established Terrier IR platform for basic text indexing and retrieval, its salient utility comes from its expressive Python operators, which allow for different IR operations to be combined in different flexible ways. Each operation applies a transformation upon a dataframe, while operators are defined with clear semantics in relational algebra. Going further, we have recently included additional support for BERT-based text re-rankers (such as EPIC) and dense retrieval implementations (such as ANCE and ColBERT). Transformer pipelines can be tuned and evaluated in a declarative manner. To increase the reusability of this framework as a resource for the IR community, PyTerrier provides easy access to a variety of standard benchmark datasets, including pre-built indices. Finally, we highlight the advantages of such a framework for information retrieval researchers and educators.

BibTeX @inproceedings{macdonald:cikm2021-pyterrier, author = {Macdonald, Craig and Tonellotto, Nicola and MacAvaney, Sean and Ounis, Iadh}, title = {PyTerrier: Declarative Experimentation in Python from BM25 to Dense Retrieval}, booktitle = {CIKM}, year = {2021} }