关键词: 数据搜索, 数据意义建构, 生成式智能搜索引擎, 智能化科研, 数据密集型科学

Abstract: Generative intelligent search engines, which integrate generative artificial intelligence technology and retrieval techniques, have the potential to support researchers in overcoming challenges during data sensemaking, but they also come with certain risks. Drawing upon sense-making theory, information behavior model, information search process model, and information search behavior model, this study preliminarily proposes the Data Search As Data sensemaking(DS-DSM) theoretical model. To explore this model, this study conducts a user experiment on the Bing Copilot platform to understand how researchers construct the meaning of data through interaction with generative intelligent search engines. The findings indicate that in the interaction scenario of generative intelligent search engines, the essence of researchers’ data search is data sensemaking. This study also identifies the stages of this process and reveals how, at each stage, researchers cognitively, affectively, and behaviorally engage with generative intelligent search engines as a bridge to overcome gaps in the data sensemaking process to complete data-centered tasks. In data-related task situations, researchers’ sensemaking begins at the formulation stage. In contrast, in research-related task situations, some researchers first go through initiation, selection, and exploration stages without clear goals. In the formulation stage, the types of gaps researchers face and the bridges they use are the most diverse, showing the most complex behavioral, cognitive, and affective responses. Based on these findings, this study proposes practical strategies for optimizing the design of generative intelligent search engines and conducting user literacy education.

Key words: Data search, Data sensemaking, Generative intelligent search engine, Artificial intelligence for research, Data-intensive science