A recommender system by using classification based on frequent pattern mining and J48 algorithm

Abstract

User's behavior modeling on the web and extracting its patterns can be utilized for customizing search results without user's specifications. Since offering a precise suggestion to users in search engines and e-commerce is desirable for users, precision is the most important factor in such systems. The main challenge in recent researches is to improve precision and recall factors in recommender systems. In addition, classification based on frequent patterns mining is received a lot of research in data mining field. In this study a hybrid method is proposed to generate a list of interesting suggestions based on users view. To verify the precision of the proposed method, we used different classifiers. The results show that, J48 classification has the highest precision and recall for the proposed method.