考虑用户信任不对称性的朋友推荐算法

Friend Recommendation Algorithm Considering Users' Confidence Asymmetry

朋友推荐在社会生活中应用广泛。本文在分析现有问题的基础上,考虑到用户之间信任的不对称性,提出用"亲密-相似"信任度概念来刻画用户相似性。然后,本文提出了一种新的考虑用户信任不对称性的朋友推荐算法。此外,本文将新提出的推荐算法应用在了一款基于Android软件技术的拼车朋友推荐手机APP中。为此,本文设计了树状节点匹配图将地点之间的包含关系转换成树的继承结构。最后,在公用数据集Movie Lens以及本文开发的应用系统中进行了对比实验,实验结果表明新的用户相似性定义方法是有效的,并且新算法的推荐准确度较高。

The friend recommendation is popular in social life. Considering the confidence degree between users is asymmetry, a concept of 'intimacy-similarity' is proposed to describe the similarity. Then, a new friend recommendation algorithm is presented by considering uses' confidence asymmetry. Besides, new recommendation algorithm is applied to a carpool-friend recommendation system based on Android Platform. In order to consummate the practical system, a structure of item-tree is designed to transform the inclusion relationship between locations to inheritance structure of tree. Finally, some contrast experiments are tested at the public data collection of Movie Lens and the APP system. The experimental results show that the proposed algorithm is valid and achieves a better precision of the recommendation.