Search

In this paper, an efficient location-based recommender system has been proposed. The proposed method, consists of three main steps. In the first phase of proposed scheme, suggestible items are ranked on the basis of user's visiting logs in his/her profile. In the second phase of the proposed method, we have used a user attribute-based item ranking scheme to solve the problems of unvisited locations and lack of information about new users. Thus, in the second phase, most similar users to target user are detected using cosine similarity first. Then, exploration history of similar user is described as a form of directed graph. In this graph, items are described as graph nodes and user movement between locations is shown as an edge which is used for ranking nodes (items) by PageRank method. Finally, in the third phase of proposed algorithm, a weighted average of rankings vectors is calculated and most suitable items are recommended to the target user. The proposed method was evaluated using Yelp dataset and obtained results were also compared with previous methods. The results show that the hybrid approach used in the proposed method will improve the performance of the recommender systems. [ABSTRACT FROM AUTHOR]