WorkWeb system---multi-workflow management with a multi-agent system

Abstract

Current ofice systems independently manage workflows and personal schedules. However, they cannot be managed independently, because of resource constraints. The "WorkWeb System" is an expanded workflow system that is able to manage and control office resources. The "BPT agent" in the system autonomously manages each workflow process instance, trying to acquire the necessary resources to complete it in time. The WorkWeb System also provides visual interfaces to manage and control office goals and several workflow m-planning algorithms to handle exceptional cases. This paper illustrates how the resource management and work%ow dynamic scheduling are performed, and discusses issues involved with this system.

Mixed-Initiative Management of Online Calendars

“…Moreover, task U of USR3 must follow task P of USR2, e.g., because the output of task P is an input for task U; this is expressed in the net by the link U s − P e ≥ 0. The extensions needed to handle such an STN may benefit from distributed solving of the Simple Temporal Problem [22] and may involve negotiations among the schedules of different users (as partially done in [13], where a user affected by a change made by another user can accept or refuse it).…”

“…Some process management tools, such as the one described in [12], offer a complementary feature with respect of our work: they estimate the dates of execution of future tasks to help the user preview the organization of pending commitments. Other tools, such as WorkWeb System [13], schedule multiple workflows by taking the availability of actors and resources (e.g., meeting rooms) into account. The main role of the actors' personal schedulers is that of automatically (or manually) accepting or rejecting new tasks and modifications proposed by other agents.…”

What agents can do in workflow management systems

“…In these cases, agents act simply as data couriers that facilitate information exchange. Personal agents that represent users Marın and Brena 2005;Tarumi et al 1997) are a generalization of this case. A different approach is to assign worklist handling operations to control agents Madhusudan Yanli et al 2006).…”

“…In the latter category, we consider special diagnostic agents to handle exceptions , negotiations Yanli et al 2006) or voting protocols. Agents may support the system by re-planning the process Tarumi et al 1997) or simply by identifying a consistent checkpoint to resume (Madhusudan 2005).…”

“…Sophisticated features for audit, such as learning from previous experiences (Inamoto 1999), recommendation for future enactments , reputation mechanism , and adaptation to modified instances , are fairly advantaged by the features of agents. No matter the audit activity (exception handling or evaluation) we distinguish two basic update mechanisms: A bottom-up one, where agents communicate the error or the performance measure to a central entity (Madhusudan 2005;Tarumi et al 1997;Xu et al 2003;Yanli et al 2006), and a top-down mechanism, where a central entity inspects the system to identify abnormalities or collect data Ehrler et al 2006;Trappey et al 2009;Qu et al 2006). An additional interesting feature is the use of agent to agree a specific level of monitoring Wang et al 2005b) in order to reduce network traffic.…”