Network soft partition based on topological potential

Partitioning of complex networks, esp. of social networks, has been a hotly debated topic in academic circles in recent years. Since actual networks usually contain some boundary nodes that are difficult to assign to a certain community, soft partitioning is under great demand in practical applications. However, at present network partitioning is done mainly by hard partition, soft partition methods are not common. In this context, a soft partition method is proposed hereby based on topological potential and specific algorithms are also provided. This method not only considers the spread of the uncertainty of community-identity of the boundary nodes in the network, but also realizes a quantified representation of the community-identity of the boundary nodes. Experiments show that this method yields results that are consistent with those by classic methods and is more reasonable.