Determining user information goals in media-rich web sites by rethinking information scent theory

The “information goal” underlying user activity on the web is an important latent parameter. Its determination can help in a wide variety of tasks ranging from improving the quality of a search engine´s results to design and evaluation of web sites. The problem of determining user information goals is exacerbated in media-rich websites. The complexity lies in the fact that the semantics as well as the user-content interactions associated with media-rich information are context-based, emergent, and hard to capture. Furthermore, simple summary data such as traversal history or links clicked are, by themselves, insufficient to obtain insights about the information needs of a user. Doing so requires an explanatory model of user action. In this paper we propose techniques for automatically estimating user information goals from usage patterns in media-rich sites. Our research builds on ideas from information foraging theory; specifically, we propose methods to capture the influence of media-rich information and its structure on information goals and re-think the classical notion of information scent to capture the subjective perception of the value/cost/access-path of media-based informational entities. Putative information goals can subsequently be estimated from browsing patterns. Experimental investigations study the efficacy of the proposed techniques for estimating user goals and user-content interaction patterns in complex media rich sites.