摘要
在基于位置的社交网络中,好友推荐主要是基于用户共同好友数量、用户行为偏好的相似性实现,而位置推荐则主要是基于地理位置进行空间聚类、用户间最长公共访问子序列实现,但目前推荐方法对用户行为偏好描述缺少语义活动信息支持,刻画用户间的关系也缺乏必要的个体信任关系描述,同时尚未综合利用第三方社交网应用中对位置的大众评分及个人评分,因而导致推荐质量不高.针对此问题,在综合考虑用户语义活动偏好、社交信任、位置合成评分以及物理距离等因素的前提下,提出FRBTA和LRBTA算法分别进行好友和位置推荐.实验结果表明,本文提出的推荐算法是可行且有效的.
In LBSNs ( Location-based Social Networks ), friend recommendation results are mainly decided by the number of common friends or depending on similar user preferences. And recommended location are basically implemented by clustering geographic locations in spatial space or by accessing the longest common sub-sequences. However, lack of description of semantic information about user activity preferences, insufficiency in building social trust among users relationships and individual score ranking by a crowd or the person from third party of social networks make recommendation quality undesirable. Aiming at this issue, two algorithms, FRBTA and LRBTA, are proposed in this paper to recommend best friends and locations by considering multiple factors such as user semantic activity preferences, social trust, comprehensive location scores and physical distance. Experimental results show that the proposed algorithms are feasible and effective.
出处
《小型微型计算机系统》
CSCD
北大核心
2014年第10期2262-2266,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61272182
61100028
60773219)资助
关键词
LBSNs
活动相似性
社交信任
好友推荐
位置推荐
LBSNs
activity Similarity
social trust
friend recommendation
location recommendation