Technology Research On Community Detection In Complex Networks Based On Genetic Optimization

In real life,many complex systems can be described by complex networks,such as WWW,electric power system,social network and communication system all have their own complex network properties.So,analyzing complex networks can be helpful to disclose the principles and properties of complex systems.The community structure is an important property of complex networks,and one of the key points to analyze complex networks is detecting the hidden community structure.Thus,community detection,which is of great significance to scientific and practical values,has raised more and more concerns in recent years.This paper is based on genetic algorithm,and sets up a model to convert community detection problems to multi-objective optimization problems,which uses the improvement of genetic manipulation.In terms of the problems of premature convergence and slow convergence speed that caused by fixed genetic manipulation,an self-adaptive community detection algorithm that based on memetic algorithm framework is proposed.This new algorithm combines the logistic function model and genetic operators,and constructs self-adaptive operators according to the dynamic demands of genetic operators in the different phase of population evolution.The selection operator uses proportional selection strategy,and the single objective function in traditional genetic algorithm is converted into two objective functions that can reflect community properties.At the same time,an algorithm of more accurate development ability called hill climbing algorithm is used in the local search.Experiments showed this algorithm is better than the compared algorithms and can be used to detect community structure in different levels.In order to overcome the problem of non-intelligent selection of genetic algorithm in the process of iterative search,a self-adaptive immune genetic community detection algorithm is proposed,which contains an immune operator and is based on antigen affinity and antibody concentration.An immune selective probability that can more flexibly select individuals as parent population from mating pool and restrain the similar individuals in population is designed.Meanwhile,a tabu search algorithm is used in the local search which can obtain more searching chances and space.Experiments indicated that this relatively stable algorithm can better balance global dispersion search and local concentration search,improve the premature convergence problem and avoid getting stuck in the local optimal solutions.