Linked Open Data provides a valuable source of information useful for making textual data machine-readable. Such information can be tremendously useful for data-intensive applications as Recommender Systems (RS), since both the preferences of users as well as the representation of items can be improved by using such data. However, only few applications really exploit their potential power.
In this tutorial, we show how the information available in the Linked Open Data cloud can be used to develop a particular class of Recommender Systems called “semantics-aware”. We will show several methodologies to introduce semantics in recommender systems, ranging from entity linking to distributional semantics models and we will describe how such representations are used to provide users with personalized suggestions of items that can be of interest for them. Moreover, we will also sketch some preliminary work about the usage of Linked Open Data to generate personalized explanations supporting the recommendations.
The tutorial has two main objectives: 1) provide the building blocks about the semantic enrichment of textual data; 2) show how the semantic representation of items description can be exploited to build effective semantics-aware recommender systems.
During the tutorial both entity linking techniques and content-based recommender methods will be introduced.
The idea is to provide an overview of methods and techniques to build effective recommender systems relying on the semantic enrichment of textual data with a particular focus on the linked open data.