A multi-agent system product line (MAS-PL) defines an architecture, whose design and implementation is accomplished using software agents to address its common and variable features. MAS-PL promotes the large-scale reuse of common and variable agency features across multiple MAS applications. The development of MAS-PLs can be achieved through MAS-specific platforms and implementation techniques, such as conditional compilation and aspect-oriented programming (AOP). However, there is not much evidence on how these techniques provide better modularity, allowing the conception of stable MAS-PL designs. This paper presents a quantitative study on the design modularity and stability of an evolving MAS-PL. The MAS-PL was built following the reactive product line adoption approach. The product line was developed and evolved based on several versions of a conference management web-based system, named Expert Committee (EC). Our evaluation is made through a series of change scenarios related to new agency features, which are agent characteristics that enhance the system with autonomous behavior. The quantitative study consists of a systematic comparison between two different versions of the EC MAS-PL based on a MAS-specific platform, called JADE. One version was implemented with object-oriented and conditional compilation techniques. The other one relied on AOP. Our analysis was driven by well-known modularity and change impact metrics.

software product lines, multi-agent systems, software metrics, empirical software engineering