Incorporating Query Term Independence Assumption for Efficient Retrieval and Ranking using Deep Neural Networks

By Bhaskar Mitra, Corby Rosset, David Hawking, Nick Craswell, Fernando Diaz, Emine Yilmaz. 2019

31 citations recorded by the source corpus.

Classical information retrieval (IR) methods, such as query likelihood and BM25, score documents independently w.r.t. each query term, and then accumulate the scores. Assuming query term independence allows precomputing term-document scores using these models---which can be combined with specialized data structures, such as inverted index, for efficient retrieval. Deep neural IR models, in contrast, compare the whole query to the document and are, therefore, typically employed only for late stage re-ranking. We incorporate query term independence assumption into three state-of-the-art neural IR models: BERT, Duet, and CKNRM---and evaluate their performance on a passage ranking task. Surprisingly, we observe no significant loss in result quality for Duet and CKNRM---and a small degradation in the case of BERT. However, by operating on each query term independently, these otherwise computationally intensive models become amenable to offline precomputation---dramatically reducing the cost of query evaluations employing state-of-the-art neural ranking models. This strategy makes it practical to use deep models for retrieval from large collections---and not restrict their usage to late stage re-ranking.

Subfields: Artificial Intelligence, Information Systems, Management Science and Operations Research, Information Retrieval & Search

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

In Management Science and Operations Research

In Information Retrieval & Search

Citations in the corpus

ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT (2020, methodology): “DeepCT uses BERT to produce the term frequency component of BM25 in a context-aware manner, essentially representing a feasible realization of the term-independence assumptionwith neural networks [23].”
Context-Aware Document Term Weighting for Ad-Hoc Search (2020, methodology): “Although much recent work has applied contextualized neural language models like BERT [14] to IR tasks, prior work mainly focused on using BERT as black-box re-ranker [9, 27, 28, 31].”
Overview of the TREC 2019 deep learning track (2020, methodology): “However, we hope that as the body of literature on neural methods for phase 1 retrieval ( e.g. , [Boytsov et al., 2016, Zamani et al., 2018, Mitra et al., 2019, Nogueira et al., 2019]) grows, we would see a larger number of runs with deep learning as an ingredient for phase 1 in future editions of…”
Overview of the TREC 2021 Deep Learning Track (2025, unknown): “The growing percentage of “fullrank” runs may also be due to increasing application of neural methods in the full retrieval setting— either using dense retrieval methods [Lee et al., 2019] or query term independent neural ranking models [Mitra et al., 2019].”
Interpretable & Time-Budget-Constrained Contextualization for Re-Ranking (2020, methodology): “One way to make neural IR models faster at query time is to offload computation to the indexing phase, either by assuming query term independence [23] or by approximating interaction similarities [17].”
Conformer-Kernel with Query Term Independence for Document Retrieval (2020, methodology): “Document expansion based methods [Nogueira et al., 2019b,a], using large neural language models, can also be classified as part of this family of approaches, assuming the subsequent retrieval step employs a traditional QTI model like BM25.”