Abstract: At the developmental turning point brought by the generative artificial intelligence (GenAI) wave, this paper provides a comprehensive review of research on conversational search behavior. It examines how the new paradigm transforms traditional perceptions of search behavior, summarizes the current state and limitations of this field, and offers valuable research directions for future studies. By reviewing the research progress in defining the overall characteristics of human-computer interaction, modeling the behavior categories and development processes of users and agents, and characterizing the behavior features and related factors of users or agents in conversational search, it argues that the main limitations of previous studies are that research objects seldom cover the cognitive and emotional activities that drive objective dialogue actions, research scenarios are mostly limited to simulated human-human dialogue scenarios, and research topics rarely involve the evaluation of user behavior performance and the enhancement of user capabilities. Future research should inherit and develop the cognitive research paradigm in the field of interactive information retrieval, expand the research scenarios to real human-computer dialogues under the GenAI scenario, and focus on evaluating and discovering better user behavior patterns.

Key words: Conversational search, Search behavior, Behavior model, Relevant factor, Literature review