Rational Retrieval Acts: Leveraging Pragmatic Reasoning to Improve Sparse Retrieval
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913825148633088 |
|---|---|
| author | Satouf, Arthur Zenou, Gabriel Ben Piwowarski, Benjamin Boubacar, Habiboulaye Amadou Piantanida, Pablo |
| author_facet | Satouf, Arthur Zenou, Gabriel Ben Piwowarski, Benjamin Boubacar, Habiboulaye Amadou Piantanida, Pablo |
| contents | Current sparse neural information retrieval (IR) methods, and to a lesser extent more traditional models such as BM25, do not take into account the document collection and the complex interplay between different term weights when representing a single document. In this paper, we show how the Rational Speech Acts (RSA), a linguistics framework used to minimize the number of features to be communicated when identifying an object in a set, can be adapted to the IR case -- and in particular to the high number of potential features (here, tokens). RSA dynamically modulates token-document interactions by considering the influence of other documents in the dataset, better contrasting document representations. Experiments show that incorporating RSA consistently improves multiple sparse retrieval models and achieves state-of-the-art performance on out-of-domain datasets from the BEIR benchmark. https://github.com/arthur-75/Rational-Retrieval-Acts |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_03676 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Rational Retrieval Acts: Leveraging Pragmatic Reasoning to Improve Sparse Retrieval Satouf, Arthur Zenou, Gabriel Ben Piwowarski, Benjamin Boubacar, Habiboulaye Amadou Piantanida, Pablo Information Retrieval Computation and Language 68P20, 68T50 H.3 Current sparse neural information retrieval (IR) methods, and to a lesser extent more traditional models such as BM25, do not take into account the document collection and the complex interplay between different term weights when representing a single document. In this paper, we show how the Rational Speech Acts (RSA), a linguistics framework used to minimize the number of features to be communicated when identifying an object in a set, can be adapted to the IR case -- and in particular to the high number of potential features (here, tokens). RSA dynamically modulates token-document interactions by considering the influence of other documents in the dataset, better contrasting document representations. Experiments show that incorporating RSA consistently improves multiple sparse retrieval models and achieves state-of-the-art performance on out-of-domain datasets from the BEIR benchmark. https://github.com/arthur-75/Rational-Retrieval-Acts |
| title | Rational Retrieval Acts: Leveraging Pragmatic Reasoning to Improve Sparse Retrieval |
| topic | Information Retrieval Computation and Language 68P20, 68T50 H.3 |
| url | https://arxiv.org/abs/2505.03676 |