Detecting defects in software requirements specification

Abstract

This research is concerned with detecting defects in software requirements specification. Motivated by both the problem of producing reliable requirements and the limitations of existing taxonomies to provide a satisfactory level of information about defects in the requirements phase, we focus on providing a better tool for requirements analysts. Only few attempts have been made to classify defects and defect detection techniques. Scattered knowledge about defects and defect detection techniques needs compilation and re-evaluation in order to enhance the ability to discover defects in the requirements phase. Toward this end, this work presents a taxonomy of requirements defects and the causes of their occurrences. The purpose is to reach a comprehensive understanding of both the sources of the problem and the solutions of possible defects and defect detection techniques. The taxonomy's design is based on the analysis of each defect and its sources. In addition, this paper proposes a combined-reading technique for defects in requirements. The proposed technique avoids the shortcomings of other reading techniques. The result of applying the recommendations of this work specifically improves the quality of the requirements specification and generally software quality.