Introducing Executable Product Models for the Service Industry

In these days, companies need highly flexible processes to deal with changing environments and competitors. Using traditional process models that have to be modeled in all details in advance, one gives up a considerable part of this flexibility. This paper presents a new approach for the information-intensive service industry, the executable product model (EPM), based on work done by van der Aalst et al. [2001] that does not show this shortcoming. In an EPM, information dependencies are represented as graph. Implicitly, this graph models all possible processing sequences and alternative processing variants. An algorithm is presented that can execute EPMs preserving all information dependencies. A multi-agent system with two different negotiation protocols is used for resource allocation. A discrete event simulation shows the principal proof-of-concept. For several structurally different EPMs, the efficiency of both protocols is statistically compared