SERUM: Collecting Semantic User Behavior for Improved News Recommendations
We present our recent work on recommending personalized news articles to users based on implicit collected feedback and large scale semantic datasets. Our personalized recommendation application SERUM exploits the fact that semantically linked and structured infor- mation becomes more and more available driven by a strong research community. Our solution combines these semantic, encyclopedic knowl- edge sources with a large news article dataset and collected implicit user feedback using an RDFa based Web application. In a first step, we com- pute semantically related entities of interest, such as similar artists or genres, based on a user behavior model using graph-based algorithms. In a second step, we utilize these interest entities for computing news arti- cle recommendations. An RDFa schema has been designed that enables standard annotations in any XHTML Web-page, thus making structured data available for the adaptation process, but also for any service or tool that supports the RDFa standard.