Knowledge-Assisted Optimization Model Formulation - Unik-Opt

In this paper, we describe a knowledge-assisted optimization model formulation system UNIK-OPT (UNIfied Knowledge-OPTimization). To envision the desirable features of UNIK-OPT, we first establish the design criteria of knowledge-assisted modeling systems. The most distinctive criterion pursued in this research is the independent management of knowledge base from multiple optimization models. To achieve these criteria, we articulate four levels of modeling views: semantic view, modeling language view, mathematical notational view and tabular view. In semantic view, the associations between attributes, blocks of terms and constraints are represented in a constraint network, and a block of terms is represented as a pair of coefficient and variable. Thus, the formulation reasoning is esteemed as a process of helping the construction of a specific semantic model by adding user's problem definition to the extracted relevant semantic level modeling knowledge. Then the specific semantic model can be transformed into other views of model automatically. The prototype UNIK-OPT is developed to realize this idea, and is illustrated with a refinery plant.