Towards Social-based User Modeling and Personalization
Michal Barla · 2010
As the Web grows and the amount of available information increases, new problems such as information overload or lost-in-hyperspace problem emerge. The solution is to shift from “one-size-fits-all ” approach and to provide personalized surfing experience, which would take into account differences among users. These differences are captured in a user model, a structure holding all relevant user-related information. We present contributions in both data collection and processing stages of user modeling process, focusing on open corpus web-based systems, where the content can be dynamically added or changed and we have no content available in the design stage of the web-based system. We introduce two methods falling within the scope of data collection stage, a method for comprehensive logging of user activity on the Web with preserved semantics, which combines client side and server side logging into a stream of user events with clearly defined meaning, and a method for capturing logs of “wild ” Web surfing based on a specialized proxy sever. The second group of methods, devoted to actual user model creation within an open corpus environment consists of a method for user model inference based on rules expressing navigational patterns, a method for term-based open corpus user modeling, which can be applied to capture user’s interest across the third-party web-sites and web-based systems. We proposed also a method for finding relations between terms based on folksonomies, which supports our term-based user modeling approach. The proposed methods were evaluated by means of software tools that were incorporated in research projects aimed Recommended by thesis supervisor: