Graph evolution via social diffusion processes

摘要: We present a new stochastic process, called as Social Diffusion Process (SDP), to address the graph modeling. Based on this model, we derive evolution algorithm and series of graphbased approaches solve machine learning problems, including clustering semi-supervised learning. SDP can be viewed special case Matthew effect, which is general phenomenon in nature societies. use social event metaphor intrinsic process for broad range data. evaluate our large number frequently used datasets compare other state-of-the-art techniques. Results show that outperforms existing methods most cases. also applying into functionality analysis microRNA discover biologically interesting cliques. Due availability graph-based data, model potentially have applications wide range.