The development of an ontology-based adaptive personalized recommender system

Recommender systems have changed the way we live for they provide strategies that help users search or make decisions within the overwhelming information spaces nowadays. They have played an important role in some area such as e-commerce and e-learning. Most of the recommender systems suffer from different weaknesses introduced by different technique, such as the New Item Problem, the Rating Sparsity Problem, the Limited Content Analysis Problem, the Overspecialization Problem, the Suggestion Ability Static Problem and the New User Problem. The strategy for personalization is to filter recommended items from the available items according to the user's preferences. And deficiency exists in personalized recommendation; that is the quality may be reduced because the user's preferences narrow down the searching spaces too much causing the Overspecialization Problem. In this paper we propose a hybrid recommendation strategy of content-based and knowledge-based, in which aiming to solve the three problems mentioned above. And we also come out with a personalization strategy that overcomes the Overspecialization Problem.