Task-Oriented Query Reformulation with Reinforcement Learning
Open accessThe DOI matches its Crossref record; 26 of the 26 references it cites were found in a registry or an index.
Abstract
Search engines play an important role in our everyday lives by assisting us in finding the information we need.When we input a complex query, however, results are often far from satisfactory.In this work, we introduce a query reformulation system based on a neural network that rewrites a query to maximize the number of relevant documents returned.We train this neural network with reinforcement learning.The actions correspond to selecting terms to build a reformulated query, and the reward is the document recall.We evaluate our approach on three datasets against strong baselines and show a relative improvement of 5-20% in terms of recall.Furthermore, we present a simple method to estimate a conservative upperbound performance of a model in a particular environment and verify that there is still large room for improvements.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Hochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318 · cited 4×
“RL-CNN and RL-RNN are the models described in Section 2.1, in which the former uses CNNs to encode query and term features and the latter uses RNNs (more specifically, bidirectional LSTMs).”
MethodInfluentialAdam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · cited 2×
“For the optimizer, we use ADAM [8] with α = 10−4, β1 = 0.”
Learning representations by back-propagating errors
Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986) · Nature · doi:10.1038/323533a0 · Cited by 31,892 · cited 2×
“We minimize C a and C b using stochastic gradient descent (SGD) with the gradient computed by backpropagation (Rumelhart et al., 1988).”
MethodEfficient Estimation of Word Representations in Vector Space
Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2013) · arXiv · doi:10.48550/arxiv.1301.3781 · Cited by 26,568 · cited 3×
“We use pretrained embed-dings from Mikolov et al. (2013), and it contains 374,000 words.”
MethodInfluentialMastering the game of Go with deep neural networks and tree search
Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., van den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., et al. (2016) · Nature · doi:10.1038/nature16961 · Cited by 15,612
Convolutional Neural Networks for Sentence Classification
Kim, Y. (2014) · Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.3115/v1/d14-1181 · Cited by 14,831 · cited 2×
“We convert the sequence {vj } to a fixed-size vector φa (v) by using either a Convolutional Neural Network (CNN) followed by a max pooling operation over the entire sequence [7] or by using the last hidden state of a Recurrent Neural Network (RNN).”
MethodWordNet: a lexical database for English
Miller, G. A. (1995) · Communications of the ACM · doi:10.1145/219717.219748 · Cited by 14,015
Simple Statistical Gradient-Following Algorithms for Connectionist Reinforcement Learning
Williams, R. J. (1992) · Machine Learning · doi:10.1007/bf00992696 · Cited by 3,416
5.7 The WORDNET lexical database of English
(2019) · Computer Corpus Lexicography · doi:10.1515/9781474471459-050 · Cited by 1,865
A Study of Smoothing Methods for Language Models Applied to Ad Hoc Information Retrieval
Zhai, C., & Lafferty, J. (2017) · ACM SIGIR Forum · doi:10.1145/3130348.3130377 · Cited by 1,537
Relevance-Based Language Models
Lavrenko, V., & Croft, W. B. (2017) · ACM SIGIR Forum · doi:10.1145/3130348.3130376 · Cited by 1,422 · cited 5×
“A popular instantiation of a local method is the relevance model, which incorporates pseudo-relevance feedback into a language model form [10].”
MethodInfluentialQuary Expansion Using Local and Global Document Analysis
Xu, J., & Croft, W. B. (2017) · ACM SIGIR Forum · doi:10.1145/3130348.3130364 · Cited by 1,252 · cited 1×
“It typically expands the original query by adding terms from, for instance, dictionaries of synonyms such as WordNet (Miller, 1995), or from the initial set of retrieved documents (Xu and Croft, 1996).”
BackgroundConvolutional Neural Networks for Sentence ClassificationPreprint
Kim, Y. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1408.5882 · Cited by 1,112 · cited 2×
“We convert the sequence {vj } to a fixed-size vector φa (v) by using either a Convolutional Neural Network (CNN) followed by a max pooling operation over the entire sequence [7] or by using the last hidden state of a Recurrent Neural Network (RNN).”
MethodRelevance based language models
Lavrenko, V., & Croft, W. B. (2001) · Proceedings of the 24th annual international ACM SIGIR conference on Research and development in information retrieval · doi:10.1145/383952.383972 · Cited by 1,037
A study of smoothing methods for language models applied to Ad Hoc information retrieval
Zhai, C., & Lafferty, J. (2001) · Proceedings of the 24th annual international ACM SIGIR conference on Research and development in information retrieval · doi:10.1145/383952.384019 · Cited by 889 · cited 3×
“We use a Dirichlet smoothed language model [21] to compute a language model from a document d ∈ D0:”
MethodInfluentialQuery expansion using local and global document analysis
Xu, J., & Croft, W. B. (1996) · Proceedings of the 19th annual international ACM SIGIR conference on Research and development in information retrieval - SIGIR '96 · doi:10.1145/243199.243202 · Cited by 831 · cited 2×
“It typically expands the original query by adding terms from, for instance, dictionaries of synonyms such as WordNet (Miller, 1995), or from the initial set of retrieved documents (Xu and Croft, 1996).”
BackgroundSimple statistical gradient-following algorithms for connectionist reinforcement learning
Williams, R. J. (1992) · Machine Learning · doi:10.1023/a:1022672621406 · Cited by 817
Selecting good expansion terms for pseudo-relevance feedback
Cao, G., Nie, J.-Y., Gao, J., & Robertson, S. (2008) · Proceedings of the 31st annual international ACM SIGIR conference on Research and development in information retrieval · doi:10.1145/1390334.1390377 · Cited by 426 · cited 4×
“Supervised Learning (SL): Here we detail a deep learning-based variant of the method proposed by Cao et al. (2008).”
MethodInfluentialAnalyzing and evaluating query reformulation strategies in web search logs
Huang, J., & Efthimiadis, E. N. (2009) · Proceedings of the 18th ACM conference on Information and knowledge management · doi:10.1145/1645953.1645966 · Cited by 277 · cited 2×
“For instance, Huang and Efthimiadis (2009) estimate that 28-52% of all the web queries are modifications of previous ones.”
BackgroundQuery Expansion Using Word Embeddings
Kuzi, S., Shtok, A., & Kurland, O. (2016) · Proceedings of the 25th ACM International on Conference on Information and Knowledge Management · doi:10.1145/2983323.2983876 · Cited by 186 · cited 2×
“[9], the top-N terms are selected based on the cosine similarity of their embeddings against the original query embedding.”
BackgroundMethodImproving Information Extraction by Acquiring External Evidence with Reinforcement Learning
Narasimhan, K., Yala, A., & Barzilay, R. (2016) · Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d16-1261 · Cited by 146 · cited 2×
“Perhaps the closest work to ours is that by Narasimhan et al. (2016), in which a reinforcement learning based approach is used to reformulate queries iteratively.”
BackgroundMethodAsk the Right Questions: Active Question Reformulation with Reinforcement Learning
Buck, C., Bulian, J., Ciaramita, M., Gajewski, W., Gesmundo, A., Houlsby, N., & Wang, W. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1705.07830 · Cited by 123 · cited 1×
“To solve these problems, we extend the model to sequentially generate a reformulated query, as proposed by Buck et al. (2017).”
MethodUsing Word Embeddings for Automatic Query ExpansionPreprint
Roy, D., Paul, D., Mitra, M., & Garain, U. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1606.07608 · Cited by 87 · cited 1×
“Embeddings Similarity: Inspired by the meth-ods proposed by Roy et al. (2016) and Kuzi et al. (2016), the top-N terms are selected based on the cosine similarity of their embeddings against the original query embedding.”
MethodLearning Concept Embeddings for Query Expansion by Quantum Entropy Minimization
Sordoni, A., Bengio, Y., & Nie, J. (2014) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v28i1.8933 · Cited by 51 · cited 2×
“An alternative technique based on supervised learning is to learn a common latent representation of queries and relevant documents terms by us-ing a click-through dataset (Sordoni et al., 2014).”
MethodImproving Information Extraction by Acquiring External Evidence with Reinforcement LearningPreprint
Narasimhan, K., Yala, A., & Barzilay, R. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1603.07954 · Cited by 34 · cited 2×
“Perhaps the closest work to ours is that by Narasimhan et al. (2016), in which a reinforcement learning based approach is used to reformulate queries iteratively.”
BackgroundMethodDíaz, F. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-30671-1_38 · Cited by 18 · cited 2×
“This can be viewed as navigating a graph where the nodes represent queries and associated retrieved results and edges exist between nodes whose queries are simple reformulations of each other [3].”
Background
Research works printed but not linked: 2
- [10]10. Victor Lavrenko and W Bruce Croft. 2001. Rele- vance based language models. In Proceedings of the 24th annual international ACM SIGIR confer- ence on Research and development in information retrieval, pages 120-127. ACM.
- [14]14. Rodrigo Nogueira and Kyunghyun Cho. 2016. End- to-end goal-driven web navigation. In Advances in Neural Information Processing Systems, pages 1903-1911.
Cited by
Liu, P., Yuan, W., Fu, J., Jiang, Z., Hayashi, H., & Neubig, G. (2023) · ACM Computing Surveys · doi:10.1145/3560815 · Cited by 4,361
“Discrete Prompt Search [52, 125] Query reformulation [90, 90] Reformulate the input into a query form Query reformulation commonly focuses on information extraction and question answering tasks, while prompt learning can be applied to a variety of NLP tasks”
BackgroundLiu, P., Yuan, W., Fu, J., Jiang, Z., Hayashi, H., & Neubig, G. (2021) · arXiv · doi:10.48550/arxiv.2107.13586 · Cited by 4,352
“Discrete Prompt Search [52, 125] Query reformulation [90, 90] Reformulate the input into a query form Query reformulation commonly focuses on information extraction and question answering tasks, while prompt learning can be applied to a variety of NLP tasks”
BackgroundA Gentle Introduction to Reinforcement Learning and its Application in Different Fields
Naeem, M., Rizvi, S. T. H., & Coronato, A. (2020) · IEEE Access · doi:10.1109/access.2020.3038605 · Cited by 257
“…games [222], language to program execution [223], extraction of information [224], sentiment analysis [225], knowledge graph reasoning [226], automatic query reformulation [227], information retrival [228], summarization [229], [230] and semantic parsing [231] where RL can contribute significantly.”
BackgroundLi, C. (2019) · Reinforcement Learning for Cyber-Physical Systems · doi:10.1201/9781351006620-6 · Cited by 188
Internet-Augmented Dialogue Generation
Komeili, M., Shuster, K., & Weston, J. (2022) · Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2022.acl-long.579 · Cited by 192
Metadata
- Authors
- Nogueira, R., & Cho, K.
- Affiliations
- New York University (US)
- Courant Institute of Mathematical Sciences (US)
- Corresponding author
- Cho, Kyunghyun
- Venue
- Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing
- Published
- 2017
- Pages
- 574-583
- OpenAlex
- W2963609889
- Cited by
- 120
- References
- 24
- Citations by year
- 2026: 3 · 2025: 1 · 2024: 5 · 2023: 10 · 2022: 15 · 2021: 22 · 2020: 24 · 2019: 20 · 2018: 17 · 2017: 3
- FWCI
- 13.33
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Conference on Empirical Methods in Natural Language Processing (EMNLP) (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Nvidia
- Coordenação de Aperfeiçoamento de Pessoal de Nível Superior
- Defense Advanced Research Projects Agency (D3M program)
- Advanced Research Projects Agency
- Topics
- Physical Sciences › Computer Science › Information Systems › Information Retrieval and Search Behavior
- Physical Sciences › Computer Science › Information Systems › Recommender Systems and Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Keywords
- query rewriting, reinforcement learning, search engines, artificial neural networks
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.