Neutrosophic Classifier-based Algorithm for Dependability Analysis in Early Phase of Software Development

Abstract

Abstract Due to the significant dependency on software, the enhancement of dependability in software attributes is the most challenging issue for software developers. The dependability attributes like: reliability, performability, security, safety, risk, maintainability perform important roles for the development of dependable software. The present study proposes a new algorithm to predict more faults efficiently in the requirement phase to assess the dependability of the software systems. Here three dependability attributes like: performability, reliability, and security, have been considered to develop the proposed model. Neutrosophic logic can be a useful tool to handle impreciseness, incompleteness of software metric values by its three independent components like: truth, indeterminate, and false components. Based on expert knowledge most important four requirement phase metrics have been considered here as input for the model development. The performance of the proposed model has been validated based on real software project data. The proposed model have been compared with different other fault prediction models by comparison criteria. The performance of the proposed model is better and it can predict more faults than other models. Software organizations can use the proposed model for the requirement phase fault count and construct dependable software.