Research On Dynamic Requirement Traceability Based On Label And Code Structure

Requirements traceability is an important step in software process management.It can help software engineers discover dependencies that exist between requirements artifacts and other software artifacts,evaluate the requirements coverage rate and calculate the influence of requirements change.Recovering and maintaining traceability automatically becomes more and more important because industrial projects become more and more complex and the number of artifacts increases fast accordingly.In recent years,extensive research efforts have been made in information retrieval(IR)based requirements traceability recovery techniques as it can recover the trace(semi)automatically.However,there are many problems,such as the difference between the requirement and the code description,the poor quality of the code annotation,and so on.Current technology is difficult to ensure the quality of traceability links.In this paper,the following work is carried out to solve the problem of recovering and optimization the traceability from requirements to code.In order to solve the problem of inconsistency between the description of requirements and code,a label-based VSM(LVSM)is proposed.First,the method uses automated methods to extract the gerund phrases from the requirements and code comments.After that,we remove gerund phrases without practical meaning.Reserved gerund phrases are used as software function labels.A label represents a system function.The label is used in the calculation of document similarity to improve the accuracy of automatically generated tracking links by reducing the redundant information in the documents.Due to the fact that there is no guarantee of the quality of the code comments,it is difficult to establish correct tracing relationship between requirements and code.In order to solve this problem,a code structure-based tracking recommendation algorithm(CSTR)is proposed in this paper.This method analyzes the code structure by abstract syntax tree(AST).And a simplified syntax tree description method is proposed to improve the efficiency of traversing the syntax tree.The code graph structure is constructed by traversing the syntax tree and stored in the graph database Neo4j.The adjacent code elements in the code graph usually have similar tracing relationships.The code graph is used to optimize the tracking links,which improves the overall quality of the traceability links.The use case and the code data in the real project iTrust are used to validate the method proposed in this paper.Experiments show that LVSM can improve the precision of traceability links and CSTR can improve the recall rate of traceability links.The other two methods,CoChaIR and Coparvo,are compared with the method in this paper,and the validity of the method is verified.