Research And Application On Software Requirements Traceability Recovery Based On Software Artifact Semantics

Maintaining the traceability of software systems is an important task in software management and development.However,in practice development,due to the pressure of time progress,traceability links are often not updated and maintained in a timely manner,and often even in missing state.Manually maintaining traceability links for software projects is time-consuming and error-prone,so this method is only suitable for small projects.In response to this problem,many studies have tried to use methods based on Information Retrieval(IR)to make this task automated.However,this method usually directly calculates the text similarity of software artifacts,and ignores the specific attributes of different software artifacts,and the returned results are often with low precision.By analyzing the problems existing in a large number of studies,this article has conducted in-depth research on how to effectively use the semantic information and key features in software artifacts to improve the traceability recovery task and proposes a hybrid approach for software traceability recovery,combines machine learning and logical reasoning to fully explore the information of software artifacts.This article focuses on traceability links from use cases to source code.On the one hand,extracting the key features of the use cases and source code based on the domain knowledge,and training machine learning the classifiers for traceability link recovery.On the other hand,exploring the relationship between artifacts,and defining a series of rules to further recover more traceability links.In particular,this article not only focuses on the structural information between source code,but also considers the relationship information between use cases.This paper significantly extends and improves traceability recovery in the following aspects.· This paper extracts the information that is essential for the software traceabilityrecovery task as the feature based on the domain knowledge,and effectively usesthe information in software artifacts to recover traceability links between softwareartifacts.· The paper converts the traceability links between software artifacts into vectorsaccording to the vital information features of software artifacts,and uses ma-chine learning classification algorithms to perform traceability recovery tasks,efficiently recovers the traceability links between software artifacts.· This paper analyzes and utilizes the structural information between source codeand the relationship information features between use cases,and defines a seriesof logical reasoning rules to capture more traceability links.We conducted a series of experiments on multiple data sets to evaluate our proposed method and compared with the related research latest methods.The results show that our method is significantly better than other related research methods.