Uncertain (Multi)Graphs for Personalization Services in Digital Libraries

Abstract

AbstractDigital Libraries organized collections of multimedia objects in a computer processable form. They also comprise services and infrastructures to manage, store, retrieve and share objects. Among these services, personalization services represent an active and broad area of digital library research. A popular way to realize personalization is by using information filtering techniques aiming to remove redundant or unwanted information from data. In this paper we propose to use a probabilistic framework based on uncertain graphs in order to deal with information filtering problems. Users, items and their relationships are encoded in a probabilistic graph that can be used to infer the probability of existence of a link between entities involved in the graph. The goal of the paper is to extend uncertain graphs definition to multigraphs and to study whether uncertain graphs could be used as a valuable tool for information filtering problems. The performance of the proposed probabilistic framework is reported when applied to a real-world domain.KeywordsRoot Mean Square ErrorRecommender SystemDigital LibraryRegular ExpressionCollaborative FilterThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.