Collective Knowledge Ontology User Profiling for Twitter -- Automatic User Profiling

摘要

How to model user interests and intentions through user profiling is an important key for providing personalized service on Internet. User profiling can be seen as the inference of user interests, intentions, characteristics, behaviors and preferences. This paper introduces a scalable and automated technique for user profiling by extracting his URLs from publicly available tweets information and using a semantic ontology in which user interests and intentions are characterized. In order to enhance the performance of our method, categorization of websites offered by OpenDNS and DBpedia collective knowledge databases are used to find the interests and intention categories of the user profile ontology. In this context, user profile ontology is populated taking these collective categories and with assertions of individuals, and relationships of interest and intention. As new concepts and relationships are defined and inferred, user profile ontology evolves continuously. Experimental results based on user's tweets confirm strongly that the proposed method improves the automatic acquisition of interests and intentions of a user profile.