Interest clustering coefficient: a new metric for directed networks like Twitter

Abstract The clustering coefficient has been introduced to capture the social phenomena that a friend of tends be my friend. This metric widely studied and shown great interest describe characteristics graph. But, is originally defined for graph in which links are undirected, such as friendship (Facebook) or professional (LinkedIn). For directed from source information consumer information, it no more adequate. We show former studies have missed much contained part graphs. In this article, we introduce new measure graphs with links, namely coefficient. compute (exactly using sampling methods) on very large graph, Twitter snapshot 505 million users 23 billion well other various datasets. additionally provide values formerly undirected metrics, first snapshot. observe higher value than classic coefficients, showing importance metric. By studying bidirectional edges also adequate while ones part. model able build random networks high finally discuss link recommendation.