Supporting Exploratory Browsing with Visualization of Social Interaction History

This thesis is concerned with the design, development, and evaluation of information visualization tools for supporting exploratory browsing. Information retrieval (IR) systems currently do not support browsing well. Responding to user queries, IR systems typically compute relevance scores of documents and then present the document surrogates to users in order of relevance. Other systems such as email clients and discussion forums simply arrange messages in reverse chronological order. Using these systems, people cannot gain an overview of a collection easily, nor do they receive adequate support for finding potentially useful items in the collection. This thesis explores the feasibility of using social interaction history to improve exploratory browsing. Social interaction history refers to traces of interaction among users in an information space, such as discussions that happen in the blogosphere or online newspapers through the commenting facility. The basic hypothesis of this work is that social interaction history can serve as a good indicator of the potential value of information items. Therefore, visualization of social interaction history would offer navigational cues for finding potentially valuable information items in a collection. To test this basic hypothesis, I conducted three studies. First, I ran statistical analysis of a social media data set. The results showed that there were positive relationships between traces of social interaction and the degree of interestingness of web articles. Second, I conducted a feasibility study to collect initial feedback about the potential of social interaction history to support information exploration. Comments from the participants were in line with the research hypothesis. Finally, I conducted a summative evaluation to measure how well visualization of social interaction history can improve exploratory browsing. The results showed that visualization of social interaction history was able to help users find interesting articles, to reduce wasted effort, and to increase user satisfaction with the visualization tool.