Social-Aware KNN Search in Location-Based Social Networks

Location-based social network services have become widely available on mobile devices. It not only helps users to strengthen their social connections, but also provides useful information. An appealing application of using these information is helping users to find proper objects(points of interests) nearby with friends' visiting experiences. In this paper, we define friend based K nearest neighbor(F-KNN) query, which aims at finding objects near the query location as well as receiving high evaluations from user's friends. To answer F-KNN query efficiently, we propose a hybrid index called F-Quadtree index, which effectively combines the geographic coordinates of objects and user's evaluation. We develop an efficient searching algorithm on the index. To further accelerate the querying process, we refine the algorithm with user based partition and memory materialization. Experimental studies on real data sets show that our methods achieve high performance.