Authors:
Omar Masmali
and
Omar Badreddin
Affiliation:
Department of Computer Science, The University of Texas, El Paso, Texas, U.S.A.
Keyword(s):
Model Driven Engineering, Software Quality Metrics, UML Class Diagram, Software Design, Code Smells.
Abstract:
Code smells and Technical debt are two common notions that are often referred to for quantifying codebase quality. Quality metrics based on such notions often reply on rigid thresholds and are insensitive to the project unique context, such as development technologies, team size, and the desired code qualities. This challenge often manifest itself in inadequate quantification of code qualities and potentially numerous false positives cases. This paper presents a novel approach that formulates code quality metrics with thresholds that are derived from software design models. This method results in metrics that, instead of adopting rigid thresholds, formulates unique and evolving thresholds specific to each code module. This paper presents the novel methodology and introduces some novel code quality formulas. To evaluate the proposed formulas, we evaluate them against open source codebase developed by experienced software engineers. The results suggest that the proposed methodology res
ults in code quality quantification that provides more adequate characterization.
(More)