基于社交圈的在线社交网络朋友推荐算法

Social Circle-Based Algorithm for Friend Recommendation in Online Social Networks

为用户推荐朋友是在线社交网络的重要个性化服务.社交网站通过用户之间是否有相同属性信息或公共邻居判断他们能否成为朋友,但由于用户注册信息不完善和对公共邻居之间关系的忽略,推荐精度不高.事实上用户的朋友可以组成多个社交圈,拥有相似社交圈的用户更易成为朋友.因此,首先提出了社交圈检测算法,进而定义用户间的社交圈相似性,基于社交圈相似程度为用户推荐新朋友.使用YouTube数据验证了该文假设;使用Facebook自我网络数据,验证了社交圈检测方法的有效性,并与3种典型检测算法比较;使用区域Facebook数据,通过与公共邻居、Jaccard相似性比较,进一步验证了朋友推荐方法的准确性.

Recommending friends to registered users is a crucial personal service of Online SocialNetworks （OSN）.OSN will recommend a friend to a user if they share some common attributesor neighbors.But the recommendation accuracy is usually not so good since users’profile infor-mation may be incomplete and the relationships between neighbors are ignored.In fact,users cangroup their friends into several social circles and two users are more likely to become friends ifthey share similar social circles.Therefore,a social circle detection algorithm is suggested atfirst,and then the social circle similarity is defined.Based on this similarity,we can recommendfriends to a user.Our hypothesis is verified by statistically analyzing the YouTube dataset.Toverify the efficiency of the social circle detection algorithm,the ego networks of Facebook areused.The experimental results show that compared with three typical detection methods,ourapproach can identify social circles efficiently and accurately.We utilize social circle similarity,common neighbor similarity andJaccard similarity to predict friend relationships in Facebook NewOrleans network.The experimental results provide strong evidence that our algorithm is moreprecise in friend recommendation.