A fuzzy traceability vector model for requirements validation

Abstract

Requirements Traceability RT is an indispensable activity to chronologically interrelate the uniquely identifiable requirements that help the developers to discover the origin of each requirement. The manual and traditional RT methods are prone to errors and overlook the concerns of analysts in retrieving the requirements. As the requirements are vague, uncertain and subjective in nature, this work combines the theory of fuzzy sets with the traditional Vector Model VM approach in modelling the vagueness, haziness and non-specificity associated with the requirements and hence facilitates tracing the requirements up to a desired degree of relevance. A number of performance measures, viz. recall, precision, F-measure, fall-outs and miss, are employed to examine the efficiency of the approach in retrieving the requirements. In addition, the proposed methodology also proposes validation metrics, viz. completeness, correctness and consistency, that result in a comprehensive, complete and consistent Software Requirements Specification SRS. The proposed approach, in this work, is exemplified in the context of a Multi-Agent System.