Semantics and Content-Based Recommendations

Content-based recommendations suggest items similar to those the user already liked in the past by building a representation of users and items based on descriptive features, which are obtained by processing textual content or low-level characteristics of the items. In this setting, the adoption of semantics-aware representations can be very useful to build a more precise representation of users and items, and, in turn, to generate better recommendations. To this end, this chapter presents a comprehensive survey of the recent trends in the area of semantics-aware content-based recommender systems. The discussion starts with an overview of historical developments, then it covers methods for semantic representations, by splitting the approaches into exogenous and endogenous representation techniques. The former rely on the integration of external knowledge sources, while the latter are based on the hypothesis that the meaning of words depends on their usage in large corpora of textual documents. Next, we discuss recent trends in the area of semantics-aware recommendations. Such trends regard novel representation methods, based on knowledge graphs and embedding techniques, as well as the exploitation of user-generated content (tags, reviews, etc.) and multimedia features. Finally, we conclude our overview by showing how content can be used to support the construction of natural language explanations and to build conversational recommender systems.