论文部分内容阅读
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 the location-based social network, friend recommendation is mainly based on the similarity of the number of user’s mutual friends and user behavior preferences, while the location recommendation is mainly based on the geographical location of the spatial clustering, the realization of the longest public access sub-sequence between users, However, the current recommendation method lacks semantic activity information support for user behavior preference description, depicts the relationship between users and lacks the necessary individual trust relationship description, and at the same time has not comprehensively utilized the public ratings and individual ratings of locations in third-party social network applications, According to this problem, FRBTA and LRBTA algorithm are put forward respectively for friend and location recommendation on the premise of considering semantic activity preference, social trust, location composition score and physical distance etc. The experimental results show that this paper proposes The recommended algorithm is feasible and effective.