Research And Application Of Graph Partition Clustering Algorithms Based On Cell-like P System

Biological computing is a inter-discipline of computer science and bio-science.Membrane computing is an important branch of biological computing.The computational models(called P system)investigated in membrane computing has the advantages of the distribution,maximum parallelism and strong fault-tolerant ability.Clustering is an important part of data mining and is also an efficient way to analysis and process data.So far,various clustering algorithms are put forward.Each of them has advantages and disadvantages and suit for different areas.Among these,the graph based partition clustering algorithms are relatively intuitive and easy to be understood and implemented.With the increasing amount of data and the increasing complexity of the data type,the traditional clustering methods cannot meet the current needs.The clustering optimal algorithm research has become a hot field.The study proposes improved graph based partition clustering algorithms to solve clustering problem firstly.Secondly it designs the new membrane system with computational universality to compute the clustering algorithm and improve the effectiveness of the algorithm.The main work is as follows:In this paper,it first introduces the research background and the current research situation of membrane computing and graph based partition clustering algorithm,the basic theory and method of membrane computing and graph theory,the research object,structure,rule and calculation Framework of P system,as well as the current graph based clustering algorithm for some improvements,and the innovation and difficulties of this study.It expands the structure and function research of P system by proposing a new model: cell-like P system with condition detection(CD-CP).The new P system is proved to be able to describe the recursive enumeration set in the state of 3 membranes.It implements the building process of a minimum spanning tree and improves the feasibility of the proposed P system in practical application.We proposed two optimized graph based partition clustering algorithm.The first one strengthens the role of structural information in a two-round minimum spanning tree clustering algorithm.And the second optimized spectral clustering algorithm with a K-NN-like method.By implementing the above improved clustering algorithms in our novel CD-CP system,we improve computational efficiency of clustering.Furthermore,we design a new method to represent text data.By mining the key frequent pattern of text database,we transform initial into vector.And we apply the improved spectral clustering algorithm to cluster the text data.It is proved that this method has a significant effect on dimension reduction and has a good performance in clustering.