SAM : Untersuchungen von Aushandlungen in Gruppen mittels Agentensimulationen

Lepperhoff, Niels

Book	PreJuSER-26266

SAM : Untersuchungen von Aushandlungen in Gruppen mittels Agentensimulationen

Lepperhoff, N. (Corresponding Author)FZJ*

2002
Forschungszentrum Jülich GmbH Zentralbibliothek, Verlag Jülich
ISBN: 3-89336-298-3

Jülich : Forschungszentrum Jülich GmbH Zentralbibliothek, Verlag, Schriften des Forschungszentrums Jülich. Reihe Informationstechnik / Information Technology 1, VI, 278 S. (2002) = Universität Dortmund, Diss., 2002

Please use a persistent id in citations: http://hdl.handle.net/2128/336

Abstract: The present study pursues three goals. First of all, it intends to Show how a simulation-based analysis can be used to investigate the basic principles of groupware and process design. An example is given by Herrmann's [1995] negotiation principle (HAP), the analysis of which represents the second goal of this study. Attention is focused on incorporating HAP into the group interaction process and an the resulting interactions. HAP is a design principle for groupware applications intended to support asynchronous, computermediated negotiations. As the third goal, an investigation is made of the extent to which a genetic algorithm facilitates the close examination of large parameter spaces. HAP is analysed with a multiagent system, SAM, specially developed for this purpose. SAM is based an various sociological theories and empirical observations. SAM permits an extensive configurability of the model in order to be able to simulate different groups and situations. The scientific value of this work is to be found in the gain in knowledge concerning group behaviour in the HAP context. Using the automaed experimenter this work Shows that no general conditions exist, which determine successful or unsuccessful groups. The success depends an specific settings of personal attributes und the group structure. SAM, moreover, represents an independent, formal scientific theory of the behaviour of groups in negotiations and may serve as a starting point for further development into a general theory of group behaviour. Furthermore, new methods and Software tools were developed in order to investigate HAP. The newly developed methods, the BDI-RC agent architecture and the automatic experimenter, and also the simulation Software for SAM represent methodological progress. Using a genetic algorithm, the automatic experimenter searches the parameter space for parameter constellations leading to the desired group behaviour. The method of using computer simulations to investigate design principles for Software and processes has proved to be viable an the whole because systematic parameter studies are possible without any disturbing influences and because influences difficult to observe empirically, such as the persons' ideas, can thus be studied. It is therefore possible to gather insights prior to an actual implementation. Expensive practical tests can be tailored to elements and situations identified as critical.

Classification: