Abstract. Many Web applications provide personalized and adapted services and contents to their users. As these Web applications are becoming increasingly connected, a new interesting challenge in their engineering is to allow the Web applications to exchange, reuse, integrate, interlink, and enrich their data and user models, hence, to allow for user modeling and personalization across application boundaries. In this paper, we present the Grapple User Modeling Framework (GUMF) that facilitates the brokerage of user profile information and user model representations. We show how the existing GUMF is extended with a new method that is based on configurable derivation rules that guide a new knowledge deduction process. Using our method, it is possible not only to integrate data from GUMF dataspaces, but also to incorporate and reuse RDF data published as Linked Data on the Web. Therefore, we introduce the so-called Grapple Derivation Rule (GDR) language as well as the corresponding GDR Engine. Further, we showcase the extended GUMF in the context of a concrete project in the e-learning domain.

Abstract. Many websites allow their users to personalize their profiles. As users subscribe to many personalization websites, such as social networks or search systems, each user owns different profiles, which are seldom compatible. Yet, there is a strong need for comparing the profiles of different users to discover shared interests, e.g., by integrating all user profiles into a global one. In this paper, we propose a novel method for integrating and ranking user interests from various profiles. Our approach relies on the identification of high-level concepts around which similar user interests are clustered. We compute the weight of each cluster with respect to the other ones, thus enabling the ranking of the most shared user interests between user profiles. 1