Alliance: A Software Framework for Software-Intensive Process Modeling, Enactment and Fuzzy Control
Résumé
This paper presents a novel approach for modelling, enacting and controlling software-intensive processes: the ALLIANCE framework (ALLIANCE stands for ALgebra and Logic for Interoperable AgeNts in Cooperative Environments). The proposed approach is based on achievements of the European PIE Project (Process Instance Evolution ESPRIT IV LTR). The paper addresses the key issue of developing a software framework that enables cooperation, decentralisation and change on-the-fly of software-intensive processes. In order to model software-intensive processes, the framework proposes a fuzzy logic based process description language based on the concept of task. To enact and control processes, the framework relies on a goal-oriented approach where software agents cooperate for achieving task goals. Software agents are autonomous (possibly decentralised) entities defined by goals they may achieve, knowledge on processes they are involved in, individual capabilities, and interaction capabilities for inter-operating with other software agents. Process quantitative control is needed to measure to which extent user-defined goals on the supported process have been reached, and if not sufficiently, how much is to be done to achieve them (how much and which resources are needed, what are consequences on the overall costs, etc.). To this end, ALLIANCE provides project managers with advanced fuzzy logic-based monitoring facilities, support for decision making and change control mechanisms. Quantitative control is considered as a step towards enterprise process improvement.