Research On Dynamic Traceability Method Between Design Documents And Source Codes In MIS

With the increasing complexity of software system,the dynamic traceability of the relationship between software artifacts(e.g.design requirements and program codes)has become more and more important.The establishment of trace links between design document and code is the key to validating software function efficiently,and it is also an important technical approach to improve the efficiency of both software development and maintenance.At present,the most widely used technology is information retrieval(IR),which identifies trace links based on textual similarities among software artifacts by constructing an IR model.Among them,Vector Space Model(VSM)is the most popular approach in dynamic traceability.However,the vocabulary mismatch problem between difference artifacts hinders the performance of IR-based approaches.Nowadays,a growing body of researchers address the issue by combining IR techniques with code dependency analysis such as method calls.However,the performance of combined approaches is highly dependent on the correctness of IR techniques and does not take full advantage of code dependency analysis.Moreover,the combined approaches do not take the importance of software system types for establishing trace links into the consideration.In the management information system(MIS),various underlying data processing usually access the database through their own SQL statements.Therefore,the functional description of the design document is usually associates with the actual SQL statement in the code.For the reasons above,this thesis combines IR techniques and code dependency analysis in the dynamic traceability between design documents and source codes of MIS,and proposes an analysis approach integrates code dependency and SQL dependency.The specific research works are as follows:(1)In order to capture and organize the code dependencies,a static analysis tool is designed and implemented to analyze the code-map's DGML files of a port cargo terminal operation system(PCTOS)based on the.NET WinForm platform so that highquality inheritance relations between two classes,contain relationships between classes and theirs methods,and call relationships between two methods can be captured.At the same time,three kinds of class-level code dependencies are abstracted from the captured method-level dependencies: class call,class inheritance,and class usage dependencies.This thesis develops a new concept,called "code dependency proximity",to measure the degree of interaction based code dependencies among classes.Then,the code dependency proximity of captured code dependencies is calculated to build a graph called Code Dependency with Proximity Graph(CDPGraph)for improving the ranking of classes in the lists of candidate links generated by IR-based techniques from design documents to code classes.(2)Due to the dependency between the design documents and the actual SQL statements of code in MIS,the functional descriptions of the design document are converted into predictive SQL,and similarity between predictive SQL and actual SQL is calculated to quantify degree of theirs association.This thesis refers to it as the "SQL dependency proximity".Based on the SQL dependency proximity,the IR-value of the class to which the actual SQL statements belong can be improved,so does the class that has a calling relationship with the class to which the actual SQL statements belong,thereby further improving the ranking of the class in the lists of candidate links.(3)In this thesis,two kinds of "proximity" above are collectively referred to as "business proximity",and a semi-automated analysis approach combining IR techniques with business proximity is proposed to recover design documents to source codes traceability.The thesis name this approach as TRIBPAnalysis(Traceability Recovery based on Information retrieval and Business Proximity Analysis)which improves the ranking of classes in the lists of candidate links through code dependency analysis and SQL dependency analysis.(4)Based on TRIBPAnalysis,a dynamic traceability prototype tool is designed and implemented.Using this tool,an experiment is conducted on trace case of PCTOS's design documents and source codes.The empirical evaluation based on PCTOS verifies that TRIBPAnalysis approach proposed in this thesis can effectively improve the accuracy of the purely IR-based traceability recovery.