Location recommendation algorithm based on temporal and geographical similarity in location-based social networks

Because the existing location recommendation algorithms in Location-based Social Networks have the characteristic of high time complexity and low recommendation accuracy, a new location recommendation algorithm based on temporal and geographical similarity is proposed by improving the traditional location recommendation algorithm in this paper. New recommendation algorithm has innovation mainly in the following three aspects: At first, new algorithm changes the traditional processing method of time dimension, it divides 24 hours into some periods of time in accordance with the time law of people's work and life, so the user similarity calculated by such periods of time will be more accurate; Secondly, the DBSCAN algorithm is improved by introducing grid thought, which makes the clustering object is no longer a single check-in point, but a grid contained a lot of check-in points, this improves the speed of recommendation algorithm greatly; Finally, a new rating function of the potential points of interest which are never visited by the user is proposed. The experimental results show that the proposed approach can improve the speed and precision of recommendation system obviously.