LAUSR

Finding desired information can still be a complex task, which is particularly challenging on specialist search engines. We propose a methodology to model search behavior evolution to better understand the familiarization process. As a case study, we analyzed features derived from search queries as well as user interface interactions of 239 users for 20 months, and following clustering, we characterized users based on their search and exploration behaviors. We analyzed the transitions between clusters over time to depict how search behavior evolution manifests. Our method enabled us to identify individuals who exhibited significant changes in search behaviors throughout their search journeys. As the study was conducted in the wild, without controlling for the tasks, topics, or demographics, the methodology holds high ecological validity for search engines that have access to unconstrained user interaction data. Ultimately, our method informs user models to better support effective web-search interactions.