Evaluation Of Software Quality Based On Support Vector Machine

With the rapid development of information technology, software products play an increasingly important role in people’s life and production and penetrated into many other industries. Software quality is the level whether the software meets the predetermined demand and is important indicators that evaluate the good or bad quality for software products. Currently, the software size is increasing and become more complex, reasonably controlling and ensuring of software product quality have become an urgent problem, The measurement evaluation of software quality characteristics is an effective means to improve software quality.Faced with the complex software systems, there are many factors that affect the software quality, the current study measure the quality of software products mainly from qualitative view. For the multiple attribute decision making problem of software quality evaluation, it should be combined quantitative with qualitative view and use scientific methods. Expansion research in this paper based on support vector machine technology, mainly from the following aspects: First, described the evaluation background and status for the software quality, found the merits and demerits, determined the significance and purpose, research methods, innovations and difficulties, laid the foundation for completing paper; second, understand the support vector machine theory, including statistical learning theory, support vector machine and kernel function theory, which the kernel function is the key for support vector machine technology; third, described the software quality theory and evaluation methods, reasonably constructed software quality evaluation index system based on the reference factors in quality model by previous study, the index system includes 7 quality attribute indexes and 22 quality sub-attributes; fourth, Constructed the mathematical model, the mathematical model is the basis for software quality evaluation, in the process of constructing SVM model, including the identification problem for kernel function and the selection of the model parameters, specifically analyzed from theoretical and experimental aspects; Fifth, example analysis for sample data, using the fitting evaluation results verify the accuracy of the software quality evaluation methods.Software quality evaluation method based on support vector machine technology researched in this paper, summarizes the advantages and shortcomings of traditional evaluation methods, and tries to build mathematical model using support vector machine technology, comparative analysis with the neural network evaluation model, fitting results show that support vector machine model evaluation method overcomes the long-time, non-uniqueness results and others demerits for neural network model, is an effective quality evaluation method.