基于社交网图和兴趣标签的协同推荐算法

为了提高用户对社交平台的粘性，通过用户的社交关系网来丰富用户的兴趣标签。以微博为例，用户的关注用户可以对用户的内容推荐进行协同性过滤，用户的关注用户的重要性受到自身粉丝数的制约，综合用户兴趣标签和用户社交网络图完成对用户推荐内容的协同过滤。以 Last．fm 数据作为测试数据集，实验结果表明：改进的算法能够较明显地提高推荐的准确度，从而表明融入用户社交关系网进行内容推荐对于提升用户的平台粘性具有一定的作用。

Research on Collaborative Recommending Algorithm based on Social Network and Interesting Tag

Through user’s social network, user’s interest tags could be showed much more enough in order that improving the dependence of platform. As an example of micro-blog, the importance of followed users would be affected by following number, the user’s recommending content would achieve the collaborative filtering both from user’s interesting tag and user’s social networking graph.The experimental results on an open dataset of Last.fm show that the improved algorithm could obviously improve the recommendation accuracy.