Intelligent Distributed User Modelling: from Semantics to Learning

Today, personalization in digital libraries and other information systems occurs separately within each system that one interacts with. However, there are several potential improvements w.r.t. such isolated approaches. Investments of users in personalizing a system either through explicit provision of information or through long and regular use are not transferable to other systems. Moreover, users have little or no control over the information that defines their profile, since user profiles are deeply buried in personalization engines. Cross system personalization, i.e. personalization that shares personalization information across different system in a user-centric way, overcomes the aforementioned problems. Information about users, which is originally scattered across multiple systems, is combined to obtain maximum leverage. Early approaches to Cross System Personalization were based on understanding user profiles semantically and transferring information between systems about the user. However, the heterogeneity in vocabulary and representation formats have made this approach unpractical; standardization efforts to create a fixed vocabulary for expressing user profiles are required to make this approach viable. Recent approaches to cross system personalization have relied on example profiles of a large number of people using multiple systems; machine learning techniques are then used to learning a mapping between profiling formats of these systems. Thus a new user is able to leverage his/her profiles from other systems to get an instant personalized experience of high quality. In this paper, we outline both of these approaches, pointing out the pros and cons of each of them, and how to use the concept of a decentralized unified user profile which acts as a Passport identifying users during their journey in information space.