ConvGQR: Generative Query Reformulation for Conversational Search

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 4998–5012 July 9-14, 2023 ©2023 Association for Computational Linguistics ConvGQR: Generative Query Reformulation for Conversational Search Fengran Mo1, Kelong Mao2, Yutao Zhu1, Yihong Wu1, Kaiyu Huang3∗, Jian-Yun Nie1∗ 1University of Montreal, Quebec, Canada 2Gaoling School of Artificial Intelligence, Renmin University of China 3Institute for AI Industry Research, Tsinghua University, Beijing, China {fengran.mo,yihong.wu}@umontreal.ca, yutaozhu94@gmail.com nie@iro.umontreal.ca, mkl@ruc.edu.cn, huangkaiyu@air.tsinghua.edu.cn Abstract In conversational search, the user’s real search intent for the current conversation turn is de- pendent on the previous conversation history. It is challenging to determine a good search query from the whole conversation context. To avoid the expensive re-training of the query encoder, most existing methods try to learn a rewriting model to de-contextualize the current query by mimicking the manual query rewrit- ing. However, manually rewritten queries are not always the best search queries. Thus, train- ing a rewriting model on them would lead to sub-optimal queries. Another useful informa- tion to enhance the search query is the potential answer to the question. In this paper, we pro- pose ConvGQR, a new framework to reformu- late conversational queries based on generative pre-trained language models (PLMs), one for query rewriting and another for generating po- tential answers. By combining both, ConvGQR can produce better search queries. In addition, to relate query reformulation to the retrieval task, we propose a knowledge infusion mech- anism to optimize both query reformulation and retrieval. Extensive experiments on four conversational search datasets demonstrate the effectiveness of ConvGQR. 1 Introduction Conversational search (Gao et al., 2022) is a rapidly developing branch of information retrieval, which aims to satisfy complex information needs through multi-turn conversations. The main challenge is to determine users’ real search intents based on the in- teraction context and formulate good search queries accordingly. Existing methods can be roughly cat- egorized into two groups. The first group directly uses the whole context as a query and trains a model to determine the relevance between the long context and passages (Qu et al., 2020; Hashemi et al., 2020; Yu et al., 2021; Lin et al., 2021b; Mao *Corresponding authors. et al., 2022a,b; Kim and Kim, 2022; Mo et al., 2023). This approach requires additional training of retriever to take the long context as input, which is not always feasible (Wu et al., 2021). What is available in practice is a general retriever (e.g., ad- hoc search retriever) that uses a stand-alone query. The second group of approaches aims at producing a de-contextualized query using query reformula- tion techniques (Elgohary et al., 2019). Such a query can be submitted to any off-the-shelf retriev- ers. We focus on this second approach. Two types of query reformulation techniques have been widely studied in the literature, i.e., query rewriting and query expansion. The former trains a generative model to rewrite the current query to mimic the human-rewritten one (Yu et al., 2020; Vakulenko et al., 2021a), while the latter focuses on expanding the current …

Content truncated to 500 words