Research Of P System In Associations Rule Mining Algorithm

Membrane computing is a new way in the area of natural computing. It is the abstract computational model which is inspired by living cells as well as the structures and functions of organs and tissues. Membrane computing modal has great parallel processing capability and also injects new ideas and methods into the computer science. At the same time, some modeling and simulations of life activities are also dependent on membrane computing model. Currently, membrane computing has attracted the attention of scholars and researchers. And it also has been used widely in the subject areas of mathematics, medicine, biology, graphics and economics, which beckons very good prospects.In recent years, with the advent of "big data" era, data mining has become a research hotspot of scholars. Association rule mining is one of the most important research topics in the field of data mining. Among them, mining frequent patterns is the main content. Frequent patterns are collections of items frequently appearing in the transaction database. In this paper, two classic frequent pattern mining algorithm(Apriori algorithm and FP-Growth algorithm) are presented as examples. Combined with in-depth understanding and analysis of these two frequent pattern mining algorithms, the parallel and distributed P systems are constructed to achieve improved algorithms respectively. Finally, the results of simulation experiments prove that P systems which implement Apriori algorithm and FP-Growth algorithm have good effectiveness and feasibility.In this paper, the first chapter is introduction, including the research background and significance of thesis topic and the current situation of membrane computing and association rule in the domestic and overseas.The second chapter is an overview of membrane computing, introducing the concept,generation and theory of membrane computing. At the same time, membrane computing model are respectively classified based on three key elements of structures, objects and evolutionary rules. Especially,a S cell-like P system is presented in detail.The third chapter is to improve traditional Apriori algorithm and to construct corresponding S cell-like P system on the basis of the implementations. This parallel P system set frequent 2 item sets into different groups according to a given certain standard. Then different groups are transmitted into different membrane to continue digging frequent patterns, reducing time complexity and improving efficiency of the algorithm.In the fourth chapter, FP-Growth algorithm is used as an example. A S cell-like P system based on the structure of Map Reduce is constructed to achieve FP-Growth Algorithm. This constructed P system has two key points.(1) All transactions are are sent into different Map and Reduce workspaces to obtain frequent 1 item sets.(2) When mining the sub-database of each frequent 1 iem, different conditional pattern bases are assigned to different Map and Reduce workspaces. Here, each Map and Reduce workspace is shown by a membrane of P system. Based on the above two points of improvement, the constructed P system can parallel complete the mining process of frequent 1 item sets and each sub-database, avoiding the need to recursively generate a lot of conditional FP-Trees in traditional FP-Growth algorithm and reducing memory space occupied. At the end, the example and simulation experiment prove the effectiveness and feasibility of this P system.The fifth chapter is the application of S cell-like P system in uer recommendation of weibo. In this part, a S cell-like P system for uer recommendation of weibo based on association rule and the structure of Map Reduce is presented, and the operation process of this P system is illustrated in detail by way of an example.The last chapter is a summary and outlook, discussing all the contents of this paper. Meanwhile, the problems that need attention and the main research directions in the future of learning are pointed out.