Visualizing Search History in Web Learning

Tetiana Tolmachova, Luyan Xu, Ivana Marenzi, Ujwal Gadiraju

Search history visualization provides a medium to organize and quickly re-find information in searching. Scientific studies show that a good visualization of a user search history should not only present the explicit activities represented by search queries and answers, but also depict the latent information exploration process in the searcher’s mind. In this paper, we propose the LogCanvasTag platform, for search history visualization. In comparison to existing work, we focus more on helping searchers re-construct the semantic relationship among their search activities. We segment a user’s search history into different sessions and use a knowledge graph to represent the searching process in each of the sessions. The knowledge graph consists of all queries and important related concepts as well as their relationships and the topics extracted from search results of each query. Sub-graphs can be extracted for each topic from the session graph for deeper insights. We also provide a collaborative perspective to support a group of users in sharing search activities and experience. Our experimental results indicate that searching experience of both independent users and collaborative searching groups benefit from this search history visualization. We present novel insights into the factors of graph-based search history visualization that help in quick information re-finding.