A platform-independent domain-specific modeling language for multiagent systems [article]

Christian Steven Hahn, Universität Des Saarlandes, Universität Des Saarlandes
2013
Hiermit versichere ich an Eides statt, dass ich die vorliegende Arbeit selbstständig und ohne Benutzung anderer als der angegebenen Hilfsmittel angefertigt habe. Die aus anderen Quellen oder indirekt übernommenen Daten und Konzepte sind unter Angabe der Quelle gekennzeichnet. Die Arbeit wurde bisher weder im In-noch im Ausland in gleicher oder ähnlicher Form in einem Verfahren zur Erlangung eines akademischen Grades vorgelegt. Saarbrücken, 31.01. Short Abstract Associated with the increasing
more » ... eptance of agent-based computing as a novel software engineering paradigm, recently a lot of research addresses the development of suitable techniques to support the agent-oriented software development. The state-of-the-art in agent-based software development is to (i) design the agent systems basing on an agent-based methodology and (ii) take the resulting design artifact as a base to manually implement the agent system using existing agent-oriented programming languages or general purpose languages like Java. Apart from failures made when manually transform an abstract specification into a concrete implementation, the gap between design and implementation may also result in the divergence of design and implementation. The framework discussed in this dissertation presents a platform-independent domain-specific modeling language for MASs called DSML4MAS that allows modeling agent systems in a platformindependent and graphical manner. Apart from the abstract design, DSML4MAS also allows to automatically (i) check the generated design artifacts against a formal semantic specification to guarantee the well-formedness of the design and (ii) translate the abstract specification into a concrete implementation. Taking both together, DSML4MAS ensures that for any well-formed design, an associated implementation will be generated closing the gap between design and code. ii iii Kurze Zusammenfassung Aufgrund wachsender Akzeptanz von Agentensystemen zur Behandlung komplexer Problemstellungen wird der Schwerpunkt auf dem Gebiet der agentenorientierten Softwareentwicklung vor allem auf die Erforschung von geeignetem Entwicklungswerkzeugen gesetzt. Stand der Forschung ist es dabei das Agentendesign mittels einer Agentenmethodologie zu spezifizieren und die resultierenden Artefakte als Grundlage zur manuellen Programmierung zu verwenden. Fehler, die bei dieser manuellen Überführung entstehen, machen insbesondere das abstrakte Design weniger nützlich in Hinsicht auf die Nachhaltigkeit der entwickelten Softwareapplikation. Das in dieser Dissertation diskutierte Rahmenwerk erörtert eine plattformunabhängige domänenspezifische Modellierungssprache für Multiagentensysteme namens DSML4MAS. DSML4MAS erlaubt es Agentensysteme auf eine plattformunabhängige und graphische Art und Weise darzustellen. Die Modellierungssprache umfasst (i) eine abstrakte Syntax, die das Vokabular der Sprache definiert, (ii) eine konkrete Syntax, die die graphische Darstellung spezifiziert sowie (iii) eine formale Semantik, die dem Vokabular eine präzise Bedeutung gibt. DSML4MAS ist Bestandteil einer (semi-automatischen) Methodologie, die es (i) erlaubt die abstrakte Spezifikation schrittweise bis hin zur konkreten Implementierung zu konkretisieren und (ii) die Interoperabilität zu alternativen Softwareparadigmen wie z.B. Dienstorientierte Architekturen zu gewährleisten. iv v Abstract Agent-based computing can be considered as promising approach and powerful technology to develop applications in complex domains by designing and developing applications in terms of autonomous software entities. Associated with the increasing acceptance of agent-based computing as a novel software engineering paradigm, recently a lot of research addresses the development of suitable techniques to support agent-oriented software development. viii ix Statement on Publications This thesis is a coherent presentation of my scientific work since June 2006. Parts of the presented material has been previously submitted, reviewed and published in various conference proceedings, book chapters and journals, and is the result of my collaboration with colleagues and the supervision of a number of master and diploma students. Part II mainly deals with the language features of DSML4MAS. Early versions of PIM4AGENTS defining the abstract syntax of DSML4MAS has been presented in 2007c ). An early version of the formal semantics appeared in 2008b,a). Chapter 5 reports on the concrete syntax and graphical editor built upon the results of the master thesis by Stefan Warwas (Warwas; 2007) and (Sadovykh et al.; . The DSML4MAS language (summarized in (Warwas and Hahn; 2009a)) won the best academic software award at the eight international conference on autonomous agents and multiagent systems. Part III mainly deals with model transformations that were partly developed during the supervision of master students Gründel (2009) and Raber (2009) . Early versions of the code generators appeared in . The internal model transformation between the interaction and behavior viewpoints of PIM4AGENTS has been presented in . Part IV deals with the evaluation of DSML4MAS. Chapter 9 reports on the industrial use case at the Saarstahl AG, which has been previously presented in 2010b,c) and . Other work published with collaborators not directly reported in this thesis-but conducted in related areas-is cited where appropriate Fischer et al.; Kahl et al.; Leon-Soto et al.; Hahn and Slomic; , 2006c Fischer et al.; 2010a; Elvesaeeter et al.; 2010d ). x xi I am indebted to Prof. Dr. Jörg Siekmann for his encouragement and motivation to conduct research in artificial intelligence. Dr. Klaus Fischer provided me with the support I needed during the ups and downs of my thesis project. He gave very constructive and illuminating feedback on drafts of this thesis. His spirit and energy have been an inspiration to me and I am grateful for the working environment he created. I am also grateful to Ingo Zinnikus, Stefan Warwas, Cristián Madrigal-Mora and my other colleagues at the MAS group of DFKI for all the discussions we had. I also thank Prof. Dr. Philipp Slusallek for letting me become a member of his research groups and I am grateful that he accepted to review my thesis
doi:10.22028/d291-26415 fatcat:e3iwb3rq7rg2rjyiv6pwgqalp4