講座內容：The objective of this talk to develop a way of enriching existing practices of system modeling, which typically result in numeric constructs, and make the existing models becoming more in rapport with real world by forming granula rmodels.

In this development, we involve one of the key paradigms of Granular Computing, namely an optimal allocation of information granularity, to deliver conceptual and algorithmic enhancements to system modeling.

Two categories of problems of allocation of information granularity are studied. In the first one, information granularity is distributed across the parameters of the existing model. In the second category of problems, information granularity becomes associated with input variables. In both cases, it is demonstrated how the resulting granular constructs augment the existing modelsand in which way they help realize a holistic way of evaluation of the original models on which their granular counterparts have been formed.

The general facet of the proposed approach is underlined; we highlight that it could be readily applied to a variety of existing models as well as invoke various formalisms of information granules (intervals, fuzzy sets, rough sets, probabilities, etc.)