案例分析：人人网的基于层次聚类的好友推荐

Case I: Friend recommendation by agglomerative hierarchical clustering

– User familiarity
• Common friends
• User profile
• User access
• User interest

Hierarchy
• Clustering to find communities in social network
– All in one community share some properties.
– These overlapping communities reveal some social
relationship of different levels.
– They help to building new friendships in the social
network.