An experimental design on the SPEM 2.0 process model element classification algorithm of the AVISPA tool through ANOVA variance analysis

Abstract

Introduction: This publication is the product of research: “SPEM 2.0 Process Model Metrics in the Reliability of its Visual Analysis” throughout 2019, which supports the work of a master’s degree in Systems Engineering at the University of Cauca.
 Objective: Rebase a process model metrics set in order to increase AVISPA reliability to support the visual analysis of SPEM 2.0 software process models.
 Methodology: In order to improve AVISPA, a systematic literature review had been performed to find software process model metrics that are potentially apt to be represented in AVISPA. Next, a set of assessments were performed in order to enhance visual analysis tool. Finally, an ANOVA statistical assessment was realized in order to find a variance differential between AVISPA versions by comparing their F1-Score process model elements classification values.
 Results: AVISPA significantly improved its general classification algorithm. Most of errors were found in SPEM 2.0 variability resolution feature and collections with duplicated elements. Multiple misclassifications still persists.
 Conclusion: General AVISPA process model elements classification is improved. However, some process model samples remain scattered according to ANOVA results.
 Originality: AVISPA is a recent solution for SPEM 2.0 software process model assessment. It's recent emergence carried to a lack of articles about software process model metrics and few works about AVISPA improvements. These are the main contributions of this paper.
 Limitations: The project has been widely expensive in terms of execution time, traceability with all software process model elements, and mainly to find experts in software process that can meet the research requirements