Structural analysis of large networks: observations and applications

摘要: Network data (also referred to as relational data, social network real graph data) has become ubiquitous, and understanding patterns in this an important research problem. We investigate how interactions networks are formed these facilitate diffusion, model behaviors, apply findings real-world problems. We examined graphs of size up 16 million nodes, across many domains from academic citation networks, campaign contributions actor-movie networks. also performed several case studies online such blogs message board communities. Our major the following: (a) discover surprising topology interactions, Popularity Decay power law (in-links a blog post decay with −1.5 exponent) oscillating connected components; (b) propose generators Butterfly generator that reproduce both established new properties found networks; (c) studies, including proposed method detecting misstatements accounting where using effects gave significant boost detection accuracy.