融合社交关系和位置影响的地点推荐算法  

Location recommendation algorithm combined with social relationships and geographic influences

作　　者：冯宇 李爱萍[1] 段利国[1] FENG Yu;LI Ai-ping;DUAN Li-guo(College of Computer Science and Technology,Taiyuan University of Technology,Jinzhong 030600,China)

机构地区：[1]太原理工大学计算机科学与技术学院,山西晋中030600

出　　处：《计算机工程与设计》2018年第9期2934-2940,共7页Computer Engineering and Design

基　　金：国家自然科学基金项目(61572345);国家科技支撑计划基金项目(2015BAH37F01)

摘　　要：为提高基于位置的社交网络服务(location-based social network service,LBSN)中地点推荐的准确率,提出一种结合社交关系和位置信息的地点推荐算法(social and location collaborative filtering,SL-CF)。以社会学六度分割理论为基础,计算对用户的信任度,获得信任用户,与相似用户融合生成邻居用户,根据融合过程中的推荐因子建立基于社交关系的预选推荐集,采用用户历史签到信息的位置影响对预选推荐列表过滤,获得推荐结果。在Foursquare数据集上的实验结果表明,该算法可以缓解数据稀疏性以及冷启动问题,验证了该算法的准确性和可行性。To improve the precision of location recommendation in location-based social network service(LBSN),a kind of location recommendation algorithm(social and location collaborative filtering,SL-CF)combined with social relationship and geographic influence was proposed.Based on sociological six degrees separation,the trusted users were obtained by calculating the trust degree of the user.Neighbor users were formed by fusing the trusted users with similar users.A preselected recommendation list based on social relations was generated according to the recommended factor came from fusion.The final recommendation result was produced by filtering preselected recommendation list according to the geographic influence of the user’s records.The algorithm can effectively relieve the data sparseness and the cold start problem may exist in the recommended process is invalidated according to the experimental results on Foursquare data sets,also the accuracy and feasibility of the algorithm are verified.

关 键 词：地点推荐 六度分割理论 信任度 位置影响 协同过滤 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]