We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.IR

Change to browse by:

cs

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Information Retrieval

Title: Evaluating the Robustness of Retrieval Pipelines with Query Variation Generators

Abstract: Heavily pre-trained transformers for language modelling, such as BERT, have shown to be remarkably effective for Information Retrieval (IR) tasks, typically applied to re-rank the results of a first-stage retrieval model. IR benchmarks evaluate the effectiveness of retrieval pipelines based on the premise that a single query is used to instantiate the underlying information need. However, previous research has shown that (I) queries generated by users for a fixed information need are extremely variable and, in particular, (II) neural models are brittle and often make mistakes when tested with modified inputs. Motivated by those observations we aim to answer the following question: how robust are retrieval pipelines with respect to different variations in queries that do not change the queries' semantics? In order to obtain queries that are representative of users' querying variability, we first created a taxonomy based on the manual annotation of transformations occurring in a dataset (UQV100) of user-created query variations. For each syntax-changing category of our taxonomy, we employed different automatic methods that when applied to a query generate a query variation. Our experimental results across two datasets for two IR tasks reveal that retrieval pipelines are not robust to these query variations, with effectiveness drops of $\approx20\%$ on average. The code and datasets are available at this https URL
Comments: Accepted for publication in the 44nd European Conference on Information Retrieval (ECIR'22)
Subjects: Information Retrieval (cs.IR)
Cite as: arXiv:2111.13057 [cs.IR]
  (or arXiv:2111.13057v1 [cs.IR] for this version)

Submission history

From: Gustavo Penha [view email]
[v1] Thu, 25 Nov 2021 12:39:51 GMT (188kb,D)
[v2] Mon, 29 Nov 2021 16:42:31 GMT (189kb,D)
[v3] Tue, 15 Feb 2022 10:25:01 GMT (188kb,D)

Link back to: arXiv, form interface, contact.