A tool-supported approach for assessing the quality of modeling artifacts

摘要

Context: Software quality engineering is increasingly gaining interests also in the Model-Driven Engineering community as testified by a large corpus of research that has been produced over the last few years. Quality models are presented as convenient artifacts to specify and organize quality attributes that are of interest for considered stakeholders.Motivation: Existing approaches enabling the specification of quality models are affected by relevant limitations including limited extensibility, artifact specificity, and manual assessment, which might lead to informal, subjective, and non-reproducible assessment processes.Goal: This paper presents an approach and related tools supporting the definition of quality models underpinning the quality assessment of modeling artifacts. Quality models are defined in terms of sets of high-level quality attributes, which are top-down decomposed into sets of subordinate attributes. An operative environment is also provided to apply the defined quality models on actual modeling artifacts enabling automated quality assessment. A set of dedicated experiments is conducted to validate the approach. The experimental results show that the proposed techniques permit modelers to define quality models taken from the literature, and apply them to assess the quality of metamodels and transformations retrieved from public repositories. The validation permitted also to analyse the performance in terms of various population structures and size.