A content recommendation framework using ontological user profiles

Abstract

In this work, a general purpose content recommendation framework using a hybrid recommendation algorithm is represented. Recommendation process is separated into different tasks; from collecting user data to representing the final recommendations. Separating these tasks to modules enables us to modify just one part of the process and compare the results. Its modular design is intended also to enable the same system be used for recommending different types of items and use different kinds of data sources. This flexibility is supported by the use of ontological user and content profiles. Different domain ontologies can be adopted to be used in the recommendation process. Items and properties of them are handled as simple ontology resources. That way, system works independent from what it is recommending as long as it has some properties that can be evaluated. So, the subject of recommendation and its properties can be in any complexity. © 2013 Springer-Verlag London.