| Multilayer complex networks are suitable models for representing high-dimensional heterogeneous systems of special importance in the era of big data.Because of the existence of inter-layer connections,the community structure in multi-layer networks has changed dramatically compared with a group of independent single-layer networks composed of the same set of nodes.For this reason,as an unsupervised learning task,community detection in multi-layer networks has become an interesting research topic of data mining and analysis in complex systems.At the same time,the task of temporal network mining is usually difficult,because not only a large number of data need to be faced,but also these data have a non-fixed nature.Usually,their organizational form and characteristics change with time.The main research work of this paper is as follows:(1)A new multilayer network community detection algorithm is proposed.This algorithm is an improved version of particle competition algorithm.The original algorithm was designed for community detection in single-layer unweighted and undirected networks.The improved version introduced in this paper can be applied to multilayer,weighted and directed networks in turn.In addition,a local measurement method is proposed to determine the optimal number of particles corresponding to the number of correctly detected communities.Computer simulation results show that the proposed technology has better performance than the existing technology.(2)A simplified representation method of time network pattern and pattern transformation is proposed.The main idea is to model each stable(persistent)state of a given time network as a community in a new static network,and to model the change of time state as a transition from one community to another.To this end,a simplified static single-layer network called target network is constructed by sampling and rearranging the original time network.The results of computer experiments on artificial simulation and real-world networks are in line with the expected purpose. |
