Information-theoretic identification of content pages for analyzing user information needs and actions on the multimedia web

Abstract

Increasing use of media in web pages is fueling the metamorphosis of the WWW into a multimedia web. The development of information organization, analysis, and search technologies that enable people to efficiently find the specific information they require becomes especially important in this context. Determining the information goal of a user is a critical step in this direction. However, the information goal is typically subjective and latent. Its determination is further complicated in multimedia websites since the semantics of media-based information is context-based and emergent. Furthermore, interaction modalities with media are, unlike static link-based browsing, complex to analyze. In this paper we address the problem of automatically determining the content pages given the browsing behavior of a user. The content pages contain the information the user came to the site to find. Thus, their identification is a critical step in reasoning about user information goals. We propose an information theoretic approach that takes into account the organization of the web site, the multimedia information content, as well as the influence of a specific browsing pattern to identify one or more pages that putatively contain the information goal(s). This method can be used irrespective of whether the user has a single information goal or is looking to satisfy multiple information needs. Experimental investigations on media rich sites illustrate the efficacy of the technique and its potential in modeling user information needs and actions in a multimedia web.