Better evaluations of Software Maintainability by using Expert Judgment and Software Measures
Assessing software maintainability is important considering the high project costs related to maintenance activities. Although expert knowledge has proven to be
effective in domains such as software engineering, the usage of expert
knowledge seems largely disassociated to code analysis and similar measurement
practices in order to assess software maintainability.
On one side, software measurement practices for determining software maintainability many times fail to address the specific needs of a project or organization due to its static nature. Software quality evaluations should be flexible and goal-driven, and this means that measures representing maintainability concepts should have different weights or priorities depending of the domain and nature of the project. On the other side, expert judgment also possess some limitations: is not always possible to find the "right" experts, and of course experts may have biases that could compromise the accuracy of the evaluations.
We believe expert judgment might support better selection and interpretation of measures for evaluating maintainability and in the other hand, providing the experts with additional information related to the systems can support more comprehensive and better software quality assessments.
There appears to be a dialectical relationship between subjective evaluation and evaluations based on product measures, and the work funded by this project attempts to build a framework for integrating both approaches for better software maintainability (and possibly other software qualities) evaluations.
Some work done along these lines is the use of Concept Mapping, which helps to structure tacit expert knowledge into a graphical representation of maintainability constructs comprised by different software measures. More info can be found here: http://simula.no/research/se/publications/Simula.SE.617
Relationship between code analysis and expert judgment according to Mantyla
