Construction And Application Of Domain Knowledge Graph For Test Cases

Software testing is an important way to ensure the reliability of software system,which runs through the whole software development process.Therefore,with the development of software system,the whole test project team will accumulate a large amount of test case data.Considering the weak semantics of test case data,it is important to use knowledge graph to help mining the semantic relationship between historical cases.Base case refers to those case templates that can express the complete business logic.In the actual testing process,testers combine specific application system components to make it a complete case that can be used for automatic testing.However,there are following problems when constructing test base cases based on domain knowledge graph.Firstly,the traditional knowledge graph constructed by domain experts has the disadvantages of high cost,limited scale and low automation.Secondly,local knowledge graphs lack of common standards because of the difference of naming standard and classification standard,which leads to the problem that knowledge graphs are unable to fusion.Thirdly,in the process of constructing test cases,it is often impossible to cover all the checkpoints.Therefore,it is important to use historical case data to complete the construction of domain knowledge graph,so as to help test project team build a new test base case,and manage the base case.To solve the above problems,this paper proposes a method to construct and apply domain knowledge graph based on test cases.Our method firstly extracts domain entities and their relationships from historical case data to construct domain knowledge graph combining with open data.Then,function-property mapping rule s and case base are generated using domain knowledge graph.Finally,based on semantic information,test base case is constructed based on functional attribute mapping rules,and test base case management and retrieval are realized based on semantics.The research contents of this thesis are listed as follows:(1)A method of automatic construction of domain knowledge graph is proposed.This method segments historical test case data with Chinese word segmentation technology,and then identifies domain entities from the segmentation results using random forest method.Finally,relations between two entities are extracted using the template matching method.(2)A multi-source knowledge graph fusion method based on open linked data is proposed.Firstly,attributes are extended using open interconnected data,then entity similarity is calculated based on graph,and entities in multi-source knowledge map are clustered and fused.In the process of case-level fusion,Markov logic network is used to coordinate knowledge conflicts.(3)A test base case construction and retrieval method based on domain knowledge graph is proposed.Function-attribute mapping rules are used to construct the test base case which specify business behavior under specific behavior and its corresponding execution results.Meanwhile,,the test case is retrieved semantically from three levels: upper level,lower level and parallel level,using the semantic relationship implied in domain knowledge graph.(4)Design and implementation of prototype system.Based on the application scenario proposed in this paper,a prototype system of domain knowledge graph construction and application oriented to bank transaction processing system testing is designed and implemented.The validity of this method is verified by demonstrating the related functions of the platform and comparing the related methods.