Implicit news recommendation based on user interest models and multimodal content analysis

This paper presents an implicit news recommender system based on the user's interests and multimodal content analysis. Multimodality is here intended as the capability of integrating multiple modes, e.g. radio-television channels and Web sites, and different media, e.g. written text and spoken content. Personal interests are inferred by natural language processing of the users' blogs. Latent semantic analysis is used to find the relationships between such interests and both online newspaper articles and broadcast news stories. The novelty of this system is the ability to treat equally and simultaneously online press reports and TV news streams. Experiments in a long-term real-world usage scenario demonstrate the quality of the proposed recommendations.