Abstract
Point-of-interest (POI) recommendation in location-based social networks (LBSNs) can solve the problem of information overload by providing personalized recommendation service, which is of great value to both users and businesses. However, the existing POI recommendation methods have not considered the effect of diversity features in check-in data, thus leading to unsatisfactory recommendation results. To address this issue, in this paper we propose an adaptive POI recommendation method (called CTF-ARA) by combining check-in and temporal features with user-based collaborative filtering. We first use probability statistical analysis method to mine user activity and similarity features of check-in behavior, variability and consecutiveness features of temporal factor. Then we use K-means algorithm to divide the users into active users and inactive users, and devise a similar user filtering algorithm based on the proposed features. Finally, we utilize cosine similarity of different time slots smoothing technique to make POI recommendation, which can operate adaptively according to the activity of user. The experimental results on Foursquare and Gowalla datasets show that CTF-ARA can improve precision and recall compared to other POI recommendation methods.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have