Mining Bug Classifier and Debug Strategy Association Rules for Web-Based Applications

摘要

The paper uses data mining approaches to classify bug types and excavate debug strategy association rules for Web-based applications. Chi-square algorithm is used to extract bug features, and SVM to model bug classifier achieving more than 70% predication accuracy on average. Debug strategy association rules accumulate bug fixing knowledge and experiences regarding to typical bug types, and can be applied repeatedly, thus improving the bug fixing efficiency. With 575 training data, three debug strategy association rules are unearthed.