This document discusses recommendation systems and how they use user and behavioral data. It provides three main points:
1) Recommendation systems gather user data from linked data, users' own websites and mobile apps, and analyze this data to learn about users and provide personalized recommendations and a more relevant service.
2) There are different recommendation methods that can be used, including collaborative filtering, content-based recommendations, knowledge-based recommendations, contextual recommendations, and social recommendations. Hybrid approaches also exist.
3) Semantic analysis and expansion of content can be used to build semantic user profiles and provide semantic recommendations. A technology called SP3/Mediatutka was presented as an example of this approach.