Analysis Of Social Networks Based On Multi-modal Particle Swarm Optimization

With the coming of Knowledge Economy Era and speedy development of Information Network Technology, the importance of social network analysis more and more obvious.In order to solve the social networks problems from many users.And better carry out Social network analysis, then finding user's interests and according to the similarity of the user community. This paper proposed a multi-modal particle swarm optimization algorithm, to solve community detection problems.Social network analysis is a complex problem;we can transform it into a multi-mode problem use mathematical theory. To address the current only solve single-modal optimization limitations,and multi-modal function is widely used in the real-world. An automatic niching particle swarm for Multi-modal Function Optimization is presented in this paper. In the proposed optimizer, niching inspired from natural ecosystem is formed automatically without any pre-specified problem dependent parameters.Results on multi-modal function experiments demonstrated that the proposed niching method is superior to the classic niching methods that with or without niching parameters, and have much higher success rate and lower computation. Further this algorithm was applied to the social network analysis to solve community detection problems,this paper proposed an adaptive community discovery algorithm based on multi-modal particle swarm optimization. Using Newman's modularity as fitness function, automatically get the number of communities in the optimization process,through Karate, Dolphins,American College Football network experiments,it shows that the proposed algorithm can effectively to predict the community and to obtain a much higher prediction accuracy.In social network analysis,users will constitute the network each node.According to user's activity or registration information, we can construct a network. This paper is conduct theoretical research and practical application based on this theory.Finally, this paper proposed personalization service platform architecture about scientific papers based on community detection. The proposed algorithm will be effective applied to this system.In practice, the system fully demonstrates the concept of user-centric.To a certain degree, it meets the individual needs of users,and establishes a good foundation for the next step development of personalization service.