Abstract: This study is aimed at developing a location recommendation system in a location based social network. The location history of the user refers to the set of Point Of Interests (POI) that the user has visited in the past. The location history of the users reflect the interests and preferences of the user in a well-deserved manner. The location history associated with the users are modeled to determine the community of the user, using an apriori based frequent POI set mining. The set of POIs visited by the members in the community and that are not visited by the user are assigned a weight calculated from the polarity and the timestamp of the tip left by the visitors from the community. Then, the POIs with highest weights are recommended to the user. The cold start problem is tackled by identifying a representative user for each community and matching the user profile with the representative user for recommendation. A comprehensive study is performed on the dataset obtained from foursquare, the popular LBSN. The empirical analysis shows that location recommendation using our proposed model gives better and accurate results when compared with the systems that use collaborative filtering or content based techniques.
K.P. Madhu and D. Manjula, 2016. A Community-Based Hybrid Location Recommendation System in Location-Based Social Networks. Asian Journal of Information Technology, 15: 1166-1174.