Research On Software Defects Based On Positive And Negative Association Rules

The problems of high credibility software are rapidly expanding with the rapid development of software technology, which has become a research hotspot in recent years. How to effectively improve the reliability of software, and to find the information they need quickly, accurately, and completely has become one urgent problem of the field of software technology. This paper carried out the research of high-reliability software defects in the motive that software defects are the direct factor to reduce the credibility of software. As an important data mining techniques, association rules reveal the hidden useful information, and solve a large extent the credibility and reliability issues. Positive and negative association rules is an important branch of association rules, which can broaden the field of research in high credibility of software defects of data mining, and have a high degree of theoretical value and comprehensive application prospects.This paper analyzed the research status of the software defects and positive and negative association rules, pointed out that further problems and study of the positive and negative association rules in the areas of application, carried out the in-depth study of software defects based on positive and negative association rules technology under the above research, and experimented through the Weka platform. The main contents of this article are as follows:(1) The latest research of data mining in the aspects of credibility of software testing, evaluation and technical for the deficiencies has been summarized, including commonly used in the test methods of defects of data mining technology, data mining systems, and software test management system. We specifically introduced the application of software defect analysis technology based on association rules in different classification of software defect(2) The common techniques of data mining and in-depth study of the association rules have been introduced. We analyzes advantages and deficiency of other techniques used in the study of software defects, focused on the relevant theory of the positive and negative association rules, and given the definition of positive and negative association rules and the calculation method of support and confidence contained with the negative project of association rules evaluation standards. (3) An interest in evaluation criteria based on the problems of the traditional framework of "suppor—confidence" has been increased, and definition of an interest and the framework of the positive negative association rule mining techniques based on the"interes—support—confidence" is given, which can mine the negative items with association rules of this technology.(4) The features, functions, and mining process of open source data mining platform of Weka are introduced. We had the re-development in the open-source environment of Weka at the same time, embedded positive and negative association rules algorithm into the Weka platform, expanded the association rules algorithm, and taken full advantage of the classes and visualization capabilities in open source of Weka. We analyzed and tested the embedded algorithm, and contrasted with the original association rules algorithm at the same time. The experiment proved that we can get a better effect of association rules from data mining platform of Weka, and the rules of which excavated are a small number of software defects, high quality, non-false rules and conflicting rules.