Exploiting modern learning algorithms for contextual recommendations
Abstract
Identifying correlations between context data, user behavior, and semantic information can lead to new services that are able to adapt to different situations. This "personalization" process can be based on recommendations on content. To better support service developers in focusing mainly on the creation of their service logic, these recommendations should be provided by a generic multipurpose recommender. Therefore, this paper proposes a generic framework that delivers "contextual recommendations" that are based on the combination of previously gathered user feedback data (i.e. ratings and clickstream history), context data, and ontology-based content categorization schemes. This paper provides a detailed overview of the specification, a short description of a possible usage scenario, and a discussion of the results.
