Task-Oriented Query Reformulation with Reinforcement Learning

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 upper-bound performance of a model in a particular environment and verify that there is still large room for improvements.

EMNLP 2017

References in corpus (2)

• Convolutional Neural Networks for Sentence Classification
• Using Word Embeddings for Automatic Query Expansion

Cited by in corpus (12)

• Deep reinforcement learning for search, recommendation, and online advertising: a survey
• Dynamic Integration of Background Knowledge in Neural NLU Systems
• Search Engine Guided Non-Parametric Neural Machine Translation
• Towards Understanding and Answering Multi-Sentence Recommendation Questions on Tourism
• Deep Reinforced Query Reformulation for Information Retrieval
• Interpretable Multi-Step Reasoning with Knowledge Extraction on Complex Healthcare Question Answering
• Learning to Coordinate Multiple Reinforcement Learning Agents for Diverse Query Reformulation
• Overcoming low-utility facets for complex answer retrieval
• Integrating Lexical and Temporal Signals in Neural Ranking Models for Searching Social Media Streams
• Analyzing Language Learned by an Active Question Answering Agent
• Improving Search through A3C Reinforcement Learning based Conversational Agent
• Subword Language Model for Query Auto-Completion