Using Semantic Web Technology and Data Mining for Personalized Recommender System to Online Shopping

Abstract

The growth of the World Wide Web provides online users a large amount of information, which has led to the emergence of web recommendation systems to recommend items, services or pages to users. Although the recommendation systems assist in retrieving and accessing interesting items automatically by mainly rely on acquiring users' historical data and matching items with the preferences of users, but still have limitations due to many issues. The issues can be identified as content overspecialization, cold start problem, and sparsity of data structures. In addition, the current recommender systems fail to make full use of the semantic information about items and the relations among them. The aim of this paper is to improve traditional recommender systems by incorporating users' information from social networks and developing users' information ontology to make personalized recommendations. Therefore, the paper proposes a framework for semantic recommender system employing user profile ontology, and products/items ontology. The proposed model includes the using of data mining techniques for knowing which classification algorithm fit well user's data analysis, which will be next using in developing the user profile ontology. Then, building a semantic application and integrate the system for checking the accuracy of the recommendation model after the validation for each techniques built. This paper is presenting the experimental results of the first step by using data mining technique of proposed model. The results show that by analyzing users' data by applying different types of classification algorithms and based on the TPR results, the Decision table algorithm gives the highest TPR by (0.871). This means that the best classifier algorithm that can be used for building user profile recommender model.