(An) exploratory search system with collaborative search path recommendation

Abstract

As increasing amount of information resource is available on the Internet, information seekers started utilizing the Internet not only to locate facts, but also to learn and to obtain the understanding of the topic they are investigating. However, search engines and web browsers which we are utilizing to search the web have been identified to be inadequate for the investigative and learning search task. Throughout this thesis, we especially valued the value of one’s information seeking history since the information contains user’s detailed interaction between the search engine and documents the person has accessed. Utilizing the search history, we implemented a prototype which is visualizing the search trails of the users followed and suggested history-based navigation approach which benefits them throughout the search process by reducing the navigational workload. Moreover, we proposed a social search concept-based recommendation method by utilizing massive search histories of a group whose goal of searching is same or similar. The user test we conducted on the history-based navigation interface revealed that the interface, in comparison with Internet Explorer, provided more reading time and improved re-accessibility by reducing the navigational workload caused by stack-based navigation approach in most commercial web browsers. Moreover, questionnaire we collected further proved that our proposed interface outperformed over Internet Explorer in many aspects, especially with properness of navigation for investigative and learning search, being informative about the status of search process, and convenience of organizing the search histories. In addition, the proposed path recommendation technique produced sets of recommendation which can be used to guide exploratory searchers to the knowledge boundary where they have not considered or to provide a shortcut to where they will eventually reach after spending some time.