An Approach to Dynamic Model Combination on Solving Decision-Making Problem

This paper discusses an approach to distributed parallel model combination for solving complicated decision-making problem. Firstly, we classify those existed models by model logic types, implement a model combination strategy to realize automatic combined model dynamic reconstruction and obtain the model series of decision-making problem solving by using and/or graph. Furthermore, during the process on model combination, we develop an index table based fast parameter matching algorithm to implement parameter match among those related models and also to apply blackboard dynamic storage technique so as to implement parameter transfer among related models. Finally, we concentrate our deep study on combined model dynamic reconstruction, fast parameter dynamic matching and transferring algorithm. Algorithm analysis listed in this paper shows that the fast parameter matching algorithm outperforms other translation methods