A framework for managing optimization models for supply chain software agents

As third party logistic services become popular, the role of software agents increases in importance in terms of the logistics scheduling of buyers and sellers. To support many models in such a portal site focused on logistics, automatic formulation and modification of optimization models embedded in the multiple software agents is necessary. The stakeholders like manufacturers and third party deliverers have their objectives and constraints in terms of delivery requirements and resource limitations. Since a variety of situations require many combinations of models, it is not easy to prepare all the necessary models in advance. To resolve this issue, we propose the primitive model approach which identifies a base model first and then modifies it to meet the modeling requirements. A prototype architecture AGENT-OPT2 is designed with the capability of rule-based model modification. This framework is demonstrated with the cooperative delivery scheduling problems.