Taking advantage of semantic-social information in recommendation systems

Abstract

With the rapid growth of Web 2.0 and the consequent success of different social network Web sites such Amazon, and MovieLens, recommender systems have recently been extensively studied by both academia and industry. In this article we combine social network analysis and semantic user profile to provide a new semantic-social recommendation, featuring a two-stage process that relies on a simple formalization of semantic user preferences that contains the user's main interests, and heuristically explores the social graph. Given a recommendation request concerning a product, the semantic-social recommendation algorithm compares the user preferences, which are found in the exploration path, with the product preferences by referencing them to domain ontology. Experiments on real-world data from Amazon, examine the quality of our recommendation method as well as the efficiency of our recommendation algorithms.