Akademik Veri Yönetim Sistemi
TÜBİTAK Uluslararası İkili İşbirliği Projesi, 2017 - 2020
Software development, required for constructing multi-agent systems (MAS) usually becomes challenging and time-consuming due to autonomy, distributedness and openness of these systems in addition to the complicated nature of internal agent behaviors and agent interactions. To facilitate MAS development, various domain-specific modeling languages (DSMLs) are proposed in the Agent-oriented Software Engineering (AOSE) research field. These DSMLs are supposed to meet software development requirements of both MAS DSML developers and MAS DSML users. Moreover, they should be evaluated to determine how they both speed up and facilitate MAS development. Although the descriptions of these languages are given in the related studies with the examples of their use, unfortunately, many of them are not evaluated either in terms of usability or quality of the generated artefacts. The evaluations in the remaining studies are made in an idiosyncratic manner without any comparison to meet developer expectations. In order to fill this gap in the AOSE research, an evaluation framework, called AgentDSM-Eval and its supporting tool, which can be used to evaluate MAS DSMLs systematically according to various quantitative and qualitative aspects of agent software development, were developed in this project. During the quantitative evaluation inside the AgentDSM-Eval tool, MAS domain coverage is determined by comparing a DSML’s metamodel with a reference MAS metamodel. In the second part of the quantitative evaluation, which is constructed on a multi-case study, the software development time and the artifact generation performance are taken into account. Finally, based on MAS developers’ feedback, the qualitative evaluation of the language is performed according to some quality metrics. Use of AgentDSM-Eval and its tool was exemplified with the evaluation of Prometheus/PDT, a well-known language in AOSE. In addition, the Analytical Hierarchy Process (AHP) based comparative evaluation method, supporting the multi-criteria decision making, was used for the evaluation of four widely used MAS DSMLs and favourite DSML for each comparison category and criteria was determined within this evaluation.