| Workflows are a major enabling technology for areas such as supply chains, logistics and production. They aim to provide computer support to the management of business processes in general. However, this technology offers little adequate support to requirements inherent to complex socio-technical systems. The domains of e-negotiations and transportation are examples that call for such systems. These domains serve us to investigate the need for an enhanced workflow technology. Hence, a clarified and a refined set of concepts and functionality for workflow management systems (WfMSs) is gathered, and appropriate solutions are proposed to deal with this set. In this context, the Workflow Reference Model is also reviewed, and an extension thereof is suggested to accommodate these concepts and functionality.; In this thesis, we study an e-negotiation support system (CONSENSUS) based on a WfMS. CONSENSUS was developed to help the user model and enact a specific kind of e-negotiation using workflows. This system requires, however, support for dynamic modification induced by unexpected events that can occur during negotiation. Current WfMSs (e.g., IBM MQ Series Workflow, BEA's WLPI) support this kind of dynamism in a limited way only, thus reducing the benefits of the CONSENSUS approach to e-negotiations.; Another complex socio-technical application, the multi-transfer container transportation (MTCT) application, exhibits inherently dynamic requirements for workflow modeling. We devise a workflow-oriented system for the processing of customer requests for container transportation. This processing is achieved by specific sequences of interdependent activities that need to be created just-in-time and then adapted to deal with unexpected events that may occur. The creation and the adaptation of activity sequences are based on an optimized resource management and activity scheduling.; In the first system, we integrate a WfMS prototype (ADEPT) that supports some of the required dynamic modifications at the workflow instance level. On one hand, this integration increases the benefits of the CONSENSUS approach. On the other hand, it sheds the light on several workflow requirements not yet fully supported by current WfMSs. In the second system, the ADEPT prototype is also used. Its API is enriched with useful functionality, and workaround solutions are required to properly cope with the definition of a workflow model and with the (dynamic) management of instances.; The realization of these two applications reaches far beyond the CONSENSUS and the MTCT projects. Indeed, the "wish list" of workflow modeling concepts and advanced functionality is carefully gathered. Particularly, we analyze best effort solutions (proper use of WfMS features, workarounds, etc.) applying three state-of-the-art WfMSs. We also work on a formal extension of a workflow meta-model to propose a correctness criterion for safely interrupting running workflow activities. This is a functionality of utmost importance. As research perspectives, the identified "wish list" and the problems expressed while experimenting with current WfMSs may define an agenda for further research in the workflow technology domain. The already investigated solutions may be applicable in the context of other challenging applications, and they may give valuable input to WfMS builders for future versions of their products. |
