ConvGQR: Generative Query Reformulation for Conversational Search
- Authors / Affiliations: Fengran Mo, Kelong Mao, Yutao Zhu, Yihong Wu, Kaiyu Huang, Jian-Yun Nie
- Published: 2023-05-25
- Source: https://arxiv.org/abs/2305.15645
- Type: Academic paper
- Topics: Conversational Search, Multi-Turn, Query Understanding
- Combining query rewriting with potential-answer generation improves conversational query quality over rewriting-only approaches.
- Jointly optimizing query reformulation and retrieval (knowledge infusion) creates a tighter link between reformulation quality and retrieval performance.
- Generative PLMs can replace manual-rewriting imitation as the basis for conversational query reformulation.
- A Survey of Conversational Search, Sep 2025, ACM
- Engineering Conversational Search Systems: A Review of Applications
- Agentic Conversational Search with Contextualized Reasoning via Reinforcement Learning
- CTR-Guided Generative Query Suggestion in Conversational Search, EMNLP
- ChatRetriever: Adapting Large Language Models for Generalized and Robust Retrieval
- CoSearchAgent: A Lightweight Collaborative Search Agent with Large Language Models
- Original paper: ConvGQR: Generative Query Reformulation for Conversational Search — arXiv:2305.15645
Background
In conversational search, the user's real search intent for the current turn depends on the previous conversation history, making it 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 learn a rewriting model to de-contextualize the current query by mimicking manual query rewriting. However, manually rewritten queries are not always the best search queries, and training a rewriting model on them limits the model's ability to produce good search queries. Another useful hint is the potential answer to the question.
Approach
ConvGQR is a framework for reformulating conversational queries based on generative pre-trained language models (PLMs). It uses two generative modules:
1. Query rewriting: rewrites the current conversational query into a standalone form that resolves coreferences and omitted context from the conversation history. 2. Potential answer generation: generates a plausible answer to the current question, providing additional semantic cues about the user's information need.
By combining the rewritten query and the potential answer, ConvGQR produces better search queries than rewriting alone.
Knowledge infusion mechanism: To relate query reformulation to retrieval performance, ConvGQR proposes a knowledge infusion mechanism that optimizes both query reformulation and retrieval jointly, aligning the reformulated queries with what the retriever actually needs.
Evaluation
Extensive experiments on four conversational search datasets demonstrate the effectiveness of ConvGQR. Specific numerical results should be verified against the original paper's tables.
Key Takeaways
Related Entries
References
Original Abstract
> In conversational search, the user's real search intent for the current turn is dependent 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 rewriting. However, manually rewritten queries are not always the best search queries. Training a rewriting model on them would limit the model's ability to produce good search queries. Another useful hint is the potential answer to the question. In this paper, we propose ConvGQR, a new framework to reformulate conversational queries based on generative pre-trained language models (PLMs), one for query rewriting and another for generating potential answers. By combining both, ConvGQR can produce better search queries. In addition, to relate query reformulation to retrieval performance, we propose a knowledge infusion mechanism to optimize both query reformulation and retrieval. Extensive experiments on four conversational search datasets demonstrate the effectiveness of ConvGQR.