Investigating search processes in collaborative exploratory web search.

摘要

People are often engaged in collaboration in workplaces or daily life due to the complexity of tasks. In the information seeking and retrieval environment, the task can be as simple as fact-finding or a known-item search, or as complex as exploratory search. Given the complex nature of the information needs, exploratory searches may require the collaboration among multiple people who share the same search goal. For instance, students may work together to search for information in a collaborative course project; friends may search together while planning a vacation.;There are demands for collaborative search systems that could support this new format of search (Morris, 2013). Despite the recognized importance of understanding search process for designing successful search system (Bates, 1990; Hearst, 2009), it is particularly difficult to study collaborative search process because of the complex interactions involved.;In this dissertation, I propose and demonstrate a framework of investigating search processes in the collaborative exploratory search. I designed a laboratory-based user study to collect the data, compared two search conditions: individual search and collaborative search as well as two task types through the study. I first applied a novel Hidden Markov Model approach to analyze the search states in the collaborative search process, the results of which provide a holistic picture of the collaborative search process. I then investigated two important components in the collaborative search process -- query behaviors and communications. The findings reveal the characteristics of query and communication patterns in the collaborative search. It also suggests that although the collaboration between two people on search did not achieve a higher performance than two individuals, the collaboration indeed make people feel more satisfied with their performance and less stressed. The results of this study not only provide implications for designing effective collaborative search systems, but also show valuable research directions and methodologies for other researchers.