Abstract
This paper presents a metadata-based framework for software architecture evaluation of quality attributes. It implements a scenario-based approach that uses dynamic analysis and code repository mining to provide an automated way to reveal degradations of scenarios on releases of web-based systems. The evaluation process has three phases: (i) dynamic analysis that collects information of scenarios in terms of measurable quality attributes; (ii) degradation analysis that processes and compares the results of the dynamic analysis in term of quality attributes for two or more existing releases of a web-based system to identify degraded scenarios considering the desired quality attributes; (iii) repository mining that looks for development issues and commits associated to code assets of the degraded scenarios. The paper also presents and discusses the obtained results of the framework instantiation for the library module of a large-scale web system. Copyright © 2014 ACM.