Research On The Algorithm Of Components Discovery In Cloud Migration

With the development of Internet technology, the traditional software has been unable to meet the needs of users and service providers, this resulting in a variety of software crisis and the decline in the quality of software development. So, during the software development, we can use the reusable components to develop software products efficiently. Software clustering is a common method of discovering the reusable components. But in the process of current development, component reuse algorithm without introducing the relationship between classes will get inaccurate software clustering results. It also can not reflect the calling relationship between modules and describe software clustering results of relevant information accurately. It makes system developers difficult to reuse the component. When dealing with large scale data, the efficiency of ordinary computing mode is low, and the results of software clustering cannot be obtained quickly. In order to solve these problems, this research proposed Pareto multi-objective genetic algorithm based on relationship types between classes for the software clustering and deploys it to the cloud. In this way, it can effectively improve the software quality and the efficiency of the component discovery. The main work is as follows:1. In the Bunch software clustering process for the component discovery, there is no module calling relation. The calling relation between modules is introduced into the process of software clustering, and applying the single-objective genetic algorithm to achieve the software clustering.2. The weight which represents the class relationship type information is introduced into the module dependency graph in order to discover the component by using Pareto multi-object software clustering algorithm.3. R-MCA (The Maximizing Cluster Approach based on the relationship between classes) and R-ECA (The Equal-Size Cluster Approach based on the relationship between classes), two different algorithms of Pareto multi-objective genetic clustering, are implemented for the component discovery in this paper. And the meanwhile, the algorithm is proposed to be deployed in the cloud to enhance the efficiency of the algorithm.