Semantic Movie Recommendations

Abstract

AbstractThe overwhelming amount of video and audio content makes it difficult for users to find new high-quality content matching the individual preferences. Recommender systems are built to suggest potentially interesting items by computing the similarity between users and items. The big challenges while creating recommender systems are the sparsity of data (the knowledge about users and items is often limited) and the popularity bias (most recommender algorithms tend to recommend popular items already known to the user). Semantic techniques supporting the graph-based representation of knowledge and the integration of heterogeneous datasets allow us to overcome these problems. The aggregation of knowledge from several different sources enables us to take into account many different aspects while computing recommendations. In addition, semantic recommender systems can provide explanations for suggested items helping the user to understand why an unknown item matches the individual user preferences. In this chapter we discuss the challenges in creating recommender systems and explain semantic approaches for the recommendation domain. We discuss the steps for building a semantic recommender system and present a semantic movie recommender system in detail. The advantages of semantic recommender systems compared to traditional recommender approaches are analyzed.KeywordsSemantic Recommender SystemIndividual User PreferencesMemory-based RecommenderInternet Movie Database (IMDb)Unique Uniform Resource Identifier (URIs)These 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.