Abstract
Requirements traceability supports many software engineering activities such as change impact analysis and requirements validation, providing benefits to the overall quality of software systems. Factors such as lack of communication, time pressure problems, and unsuccessfully implemented traceability practices result in developers losing track of requirements. Requirements traceability is a primary means to address completeness and accuracy of requirements. It is an active research topic for software engineers. Textual analysis and information retrieval techniques have been applied to the requirements traceability recovery problem for many years, due to the textual components of requirements and source code. Information retrieval techniques are semiautomatic techniques for recovering traceability links and on occasion, they have become the baseline for automatic methods applied to requirements traceability recovery. We evaluate the performance of IR techniques applied to the requirement traceability recovery process. The most popular information retrieval techniques applied to the requirements traceability recovery problem are the IR Probabilistic, Vector Space Model, and Latent Semantic Index approach. All three approaches rank documents by using one of the documents for extracting queries and the other as the documents being search using those extracted queries; however, they apply different internal logics for establishing similarities. We compared IR Probabilistic, Vector Space Model, and Latent Semantic Index approaches to evaluate their performance for requirement traceability recovery using the metrics of precision and recall. Experimental results indicate a low precision and recall for the LSI technique and high precision and low recall for both the IR probabilistic and the VSM techniques.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have