ChatRetriever: Adapting Large Language Models for Generalized and Robust Conversational Dense Retrieval
Overview
| Field | Content | |------|------| | Title | ChatRetriever: Adapting Large Language Models for Generalized and Robust Conversational Dense Retrieval | | Authors | Kelong Mao, Chenlong Deng, Haonan Chen, Fengran Mo, Zheng Liu, Tetsuya Sakai, et al. (7 authors) | | Published | 2024-04-21 | | Source | https://arxiv.org/abs/2404.13556 | | Type | Academic paper | | Section | Conversational Search |
One-line Summary
ChatRetriever adapts large language models for conversational dense retrieval via a dual-learning approach combining contrastive learning with masked instruction tuning, achieving state-of-the-art performance without requiring LLM-based query rewriting.
Background and Motivation
Conversational search requires accurate interpretation of user intent from complex multi-turn contexts. Traditional pipelines that separate retrieval, ranking, and generation struggle to meet the combined demands of natural language interaction, multi-hop reasoning, and up-to-date knowledge in the LLM era. ChatRetriever was proposed to systematize and push forward the boundaries of theory and practice in this intersection, leveraging the strong generalization capability of LLMs to robustly represent complex conversational sessions for dense retrieval.
Method
The core contribution is a simple and effective dual-learning approach:
1. Contrastive learning adapts the LLM itself for dense retrieval; 2. Masked instruction tuning on high-quality conversational instruction tuning data enhances the model's understanding of complex multi-turn sessions.
This design directly encodes conversational context into dense representations, avoiding a separate query-rewriting stage while inheriting the semantic generalization of the underlying LLM.
Experiments and Results
Per the original abstract:
- Evaluated on five conversational search benchmarks;
- Substantially outperforms existing conversational dense retrievers;
- Achieves state-of-the-art performance on par with LLM-based rewriting approaches;
- Exhibits superior robustness in handling diverse conversational contexts.
- 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
- CoSearchAgent: A Lightweight Collaborative Search Agent with Large Language Models
- ConvGQR: Generative Query Reformulation for Conversational Search
- Original paper: ChatRetriever: Adapting Large Language Models for Generalized and Robust Conversational Dense Retrieval. arXiv: https://arxiv.org/abs/2404.13556
Exact numerical results should be verified against the tables in the original PDF.
Key Takeaways for Search / Rec / Personalization
1. Architecture: adapting a retriever directly to complex inputs (e.g., conversation sessions) can rival rewriting pipelines, at lower inference cost; 2. Data: high-quality conversational instruction tuning data is critical for complex session understanding; 3. Evaluation: robustness across diverse conversational contexts matters as much as average benchmark scores; 4. Production: latency, cost, and answer trustworthiness remain hard constraints for industrial deployment.
Related Entries (Cross-references)
Original English Abstract
> Conversational search requires accurate interpretation of user intent from complex multi-turn contexts. This paper presents ChatRetriever, which inherits the strong generalization capability of large language models to robustly represent complex conversational sessions for dense retrieval. To achieve this, we propose a simple and effective dual-learning approach that adapts LLM for retrieval via contrastive learning while enhancing the complex session understanding through masked instruction tuning on high-quality conversational instruction tuning data. Extensive experiments on five conversational search benchmarks demonstrate that ChatRetriever substantially outperforms existing conversational dense retrievers, achieving state-of-the-art performance on par with LLM-based rewriting approaches. Furthermore, ChatRetriever exhibits superior robustness in handling diverse conversational contexts. Our work highlights the potential of adapting LLMs for retrieval with complex inputs like conversational search sessions and proposes an effective approach to advance this research direction.