CONQRR: Conversational Query Rewriting for Retrieval with Reinforcement Learning

Wu, Zeqiu; Luan, Yi; Rashkin, Hannah; Reitter, David; Hajishirzi, Hannaneh; Ostendorf, Mari; Tomar, Gaurav Singh

doi:10.18653/v1/2022.emnlp-main.679

Cited by 6 publications

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Common methods involve selecting relevant tokens from the search session [23,35,43] and training a generative rewriter model using human-rewritten queries paired with their respective sessions [22,26,41,50]. Some research efforts incorporate reinforcement learning [5,48] or ranking signals [27,32] to align the generation process with the downstream search task. In contrast, CDR utilizes conversational search session data to perform end-to-end dense retrieval.…”

Section: Related Work 21 Conversational Searchmentioning

confidence: 99%

“…This approach allows for the use of existing retrievers for the search process. However, it is challenging to directly optimize the rewriting towards search [21,27,32,48]. Another approach, known as conversational dense retrieval (CDR), focuses on training a conversational dense retriever to grasp the search intent by implicitly learning the latent representations of encoded queries and passages.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

ConvSDG: Session Data Generation for Conversational Search

Mo,

Yi,

Mao

et al. 2024

Companion Proceedings of the ACM on Web Conference 2024

View full text Add to dashboard Cite

Conversational search provides a more convenient interface for users to search by allowing multi-turn interaction with the search engine. However, the effectiveness of the conversational dense retrieval methods is limited by the scarcity of training data required for their fine-tuning. Thus, generating more training conversational sessions with relevant labels could potentially improve search performance. Based on the promising capabilities of large language models (LLMs) on text generation, we propose ConvSDG, a simple yet effective framework to explore the feasibility of boosting conversational search by using LLM for session data generation. Within this framework, we design dialogue/session-level and query-level data generation with unsupervised and semi-supervised learning, according to the availability of relevance judgments. The generated data are used to fine-tune the conversational dense retriever. Extensive experiments on four widely used datasets demonstrate the effectiveness and broad applicability of our ConvSDG framework compared with several strong baselines. CCS CONCEPTS• Computing methodologies → Discourse, dialogue and pragmatics; • Information systems → Information retrieval.

show abstract