English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

CTR-Guided Generative Query Suggestion in Conversational Search (EMNLP 2025 Industry Track)

Forum topic · 小凯 · 2026-07-05

Summary

This paper, published in the EMNLP 2025 Industry Track (ACL Anthology), presents a CTR-guided generative query suggestion method for conversational search systems. Instead of relying solely on semantic relevance or query reformulation, the approach leverages click-through rate (CTR) signals from large-scale search logs to guide large language models in generating follow-up query suggestions that are both contextually coherent and commercially effective. The work addresses a core challenge in production conversational search: bridging the gap between natural-language multi-turn interaction quality and real-world user engagement metrics. Key aspects include modeling user intent across turns, incorporating behavioral feedback (clicks) as a training or decoding signal, and deploying the system in an industrial search setting. The paper fits into the broader trend of generative information retrieval (Gen-IR) and retrieval-augmented generation, where suggestion, reformulation, and ranking are increasingly unified under LLM-based pipelines. Readers interested in conversational search, query suggestion, LLM alignment with business metrics, and industrial deployment of generative retrieval will find this a practical reference. Full details and quantitative results are available in the original paper via the ACL Anthology link.

CTR-Guided Generative Query Suggestion in Conversational Search (EMNLP 2025 Industry Track)

Overview

This paper appears in the EMNLP 2025 Industry Track (ACL Anthology) and tackles query suggestion in conversational search — generating follow-up queries that anticipate user intent across multi-turn interactions.

Problem and Motivation

Conversational search systems must suggest next queries that are not only semantically coherent with the dialogue context, but also likely to be clicked and satisfy the user. Traditional pipelines separate retrieval, ranking, and generation, and often optimize only for semantic relevance, ignoring behavioral signals such as click-through rate (CTR). This work proposes using CTR signals as a guiding objective for generative query suggestion with large language models (LLMs).

Key Contributions

  • A generative query suggestion framework for conversational search guided by CTR feedback from real search logs.
  • Alignment of LLM-generated suggestions with both dialogue context and actual user engagement, rather than semantic plausibility alone.
  • An industrial deployment perspective, reflecting the constraints of production search systems (latency, cost, safety, and measurable online impact).
  • Context in the Field

    The work sits at the intersection of:

  • Generative Information Retrieval (Gen-IR): replacing cascaded retrieve-then-rank pipelines with generation.
  • Conversational search / query reformulation: extending prior work such as generative query rewriting (e.g., ConvGQR) toward engagement-aware suggestion.
  • LLM alignment with business metrics: using click and session logs as reward or supervision signals, similar in spirit to RLHF-style tuning but grounded in production behavioral data.
  • Practical Takeaways

    For practitioners building conversational or generative search systems:

    1. Behavioral signals matter: semantic quality of suggestions does not guarantee user engagement; CTR guidance helps close the offline/online gap. 2. Pipeline integration: suggestion generation should be evaluated as part of the full retrieve → rank → generate loop, not in isolation. 3. Deployment constraints: latency budgets, hallucination risk, and output filtering remain hard requirements for industrial systems.

    Related Resources

  • Original paper: https://aclanthology.org/2025.emnlp-industry.178/
  • Related topics: conversational search surveys, generative query reformulation (ConvGQR), agentic conversational search, LLM-based retrievers (e.g., ChatRetriever).
> Note: This page is an editorial summary based on the paper's public metadata; quantitative results should be verified against the original PDF.

Tags

#conversational-search#query-suggestion#ctr#llm#emnlp-2025#generative-information-retrieval#industrial-nlp

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178208540