Incorporating Commitment Protocols into Tropos

• Published in: Agent-Oriented Software Engineering VI pp. 69 - 80
• Main Authors: Mallya, Ashok U., Singh, Munindar P.
• Format: Book Chapter Conference Proceeding
• Language: English
• Published: Berlin, Heidelberg Springer Berlin Heidelberg 2006; Springer
• Series: Lecture Notes in Computer Science
• Subjects: Applied sciences; Artificial intelligence; Business Process; Computer science; control theory; systems; Decomposition Tree; Exact sciences and technology; Goal Model; Interaction Protocol; Multiagent System; Multiagent system; Software architecture; Software development; Systems engineering; Artificial intelligence; Modeling; Intelligent agent; Commitment scheme; Agent oriented
• ISBN: 3540340971; 9783540340973
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/11752660_6

This paper synthesizes two trends in the engineering of agent-based systems. One, modern agent-oriented methodologies deal with the key aspects of software development including requirements acquisition, architecture, and design, but can benefit from a stronger treatment of flexible interactions. Two, commitment protocols declaratively capture interactions among business partners, thus facilitating flexible behavior and a sophisticated notion of compliance. However, they lack support for engineering concerns such as inducing the desired roles and selecting the right protocols. This paper combines these two directions. For concreteness, we choose the Tropos methodology, which is strong in its requirements analysis, but our results can be ported to other agent-oriented methodologies. Our approach is as follows. First, using Tropos, analyze requirements based on dependencies between actors. Second, select top-level protocols based on the actors’ hard goals, while respecting the logical boundaries of their interactions. Third, select refined protocols based on the actors’ soft goals. Consequently, Tropos provides a rigorous basis for modeling and composing protocols whereas the protocols help produce perspicuous designs that respect the participants’ autonomy. We evaluate our approach using a large existing case.