Information Filtering for Context-Sensitive Browsing

Browsing is one of the most popular ways to gather information in database with hypertext structure. In the browsing, a user continuously searches nodes which include useful information for her/him. Her/his interests, then, often change while the browsing. We call this type of browsing “context-sensitive browsing” in order to distinguish it from browsing with consistent interests. In this paper, we propose a method to filter the links in hypertext based on the user's browsing history. We assume that even when a user browses, following changeable interests without a clear task, the user's current interests are reflected in the content and order of nodes in the browsing history. The filtering method models user's current interests from the user's browsing history and puts the next choices in order of the nearness to the interests. We call the filtering method “context-sensitive filtering”. We have developed a browsing support system with this method for an encyclopedia in CD-ROM format. The results of an experimental evaluation, by real users, are also reported.