Test Case Reuse Based On Software Requirement

Test case reuse can significantly improve the software quality,reduce the software development cycle and make up the lack of experience for some developers.Software requirement is used to guide the designing and developing of test cases,which closely connected with software testing process.Based on the difference of software functions,software requirement can be divided into different categories,so do the test cases.In this thesis,we presented a test case reuse framework to reuse the test cases based on the software requirements.Most of the time,software requirements can be described based on text form so that software requirement categorization can be turned into text categorization.After the initialization,software requirements can be express by a vector space with high dimensionality and sparsity,which is increasing the time and space complexity in categorize process,so it may directly the effect the accuracy of requirement categorization.We analyzed the details of each categorize process and focus on the dimensionality reduction research,which is including feature extraction and feature selection.Compare with feature extraction,feature selection is easy to achieve better results with convenience approaches,so feature selection is more effectiveness for software requirement categorization.In this thesis,some traditional feature selection algorithms have been analyzed,and we presented tabu search-based feature selection algorithm to apply to software requirements categorization.Meanwhile,Genetic Algorithm(GA)has been mainly used to compare with others and comparison results showed that tabu search algorithm in feature selection process can avoid falling into the local best results and finally find the global best feature set.To verify the effectiveness and efficiency of our algorithm,we tested our tabu search-based feature selection algorithm based on Chinese Corpus.And accuracy rate,recalling rate and F1 values have been calculated to evaluate our approach.Experiment results showed that our algorithm can improve the feature selection accuracy with high efficiency in comparison with GA.At last a case study showd that tabu search-based feature selection algorithm can guarantee the accuracy of selection results in software requirement categorization.