Research On Mutation Execution Strategy For Mutation Based Fault Localization

With the development of military aviation software,abundant test cases and requirements document are accumulated and settled down.The ever-increasing volumes of information and heavy test tasks are both required a higher level of testers.In a way testers should ensure the efficiency and professionalism of the testing process as well as guarantee the quality of softwares.So that how to use historical tests data to optimization and guide future tests,has becoming to a very important and rigorous challenge.Reusing test cases can be explained to that during the process of test,people make full use of accumulated test cases and knowledges,in a view of previous studies and experiences,inspire new tests.reuse old test cases in a new developing software or in a new testing.Obviously,the reuse of test cases can give higher values to historical test cases and improve the efficiency of new testing mission.Based on the document features of military aviation software's requirements specifications(hereinafter referred to as requirement)and test cases execution record sheets(hereinafter referred to as test case),here we bring a model which has built on caculate similarity of documents to reuse requirements.Though designing reusing model,analysing the way of how judgment method of document similarity and strategy of tag extraction have influenced the degree of requirements reusing,we successfully convert the reusing problem into a Chinese long document similarity judgment and topic mining problem.By using the method that we have designed and implemented a document similarity based and test case reuse system,to standardized management historical requirements and intelligent recommend test cases.The main work and innovation points of this paper are summarized as follows:1.Proposed a new strategy of test case reuse.Compare to this,the traditional test case reuse is based on retrieval or formal modeling,in which we need either remember search word for retrieval or complicate the model when documents increased for formal modeling.Based on these limitations and characteristics of the requirements and the document,in this paper we used Natural Language Processing's document similarity method,from the perspective of document similarity to mining the similarity among requirements and the relativity between requirements and test cases.2.Researched and implemented the reuse method.We departed test case reuse in two steps,namely,similar documents recommendation and tag extraction.In the same time this paper presents the concept of structural weight by combining the word frequency weight statistical method TFIDF with the neural network word vector model doc2vec and the topic mining model LDA,on this basis,proposed our test case reuse model.3.Judging from the document similarity and tag recommendation two angles,respectively,we designed experiments,which were conducted on the sougou corpus and the real military aviation software requirements and test case document,verified the effectiveness of the reuse method.4.According to the reusing strategy and model proposed in this paper,combined with a military aviation institute of software test case reuse requirements,we designed and implemented a test case reuse system,used the methods in the actual software testing process.The result of this model works pretty well and improved the reuse efficiency of test cases.