Addressing Agent Autonomy in Business Process Management - With Case Studies on the Patient Discharge Process

摘要

This work argues for the need to represent and reason about agent autonomy in modeling and supporting business processes. We showed how to use the i~* framework to do that, recognizing that agents are allowed to pursue their goals in their own way, but also introducing constraints in the way some goals have to be achieved when necessary. Sequence of tasks can be expressed when needed [Liu 02]. A methodology presenting systematic guidelines to use i~* for requirements elicitation has been recently proposed [Cysneiros 03]. In order to diminish the gap between BPM and agent software paradigm, we propose to use the i~* framework as a front end to BPM techniques and languages. In such a way one should first use the i~* framework to understand the business and elicit requirements for the software-to-be. Once there is an established process suitable to all stakeholders, the process modelled in i~* may be mapped to BPMN [Owen 03]. Although this is future work, we have shown in section three a preliminary approach for this mapping. We also foresee that in the future, to cope with the need for flexibility, BPM may use software agents, in which case the use of i~* as a front end will be even more natural. A version of i~* called GRL [GRL] is being proposed as part of a User Requirements Notation as a standard for the ITU Telecommunication Standardization Sector (ITU-T). Future work includes studying how well the framework scales from small to very large systems.