Model Representation in Information Resource Management

Abstract

Analytical models, including mathematical programming algorithms and statistical tools, tend to be implemented as self-contained application programs in software systems. When they are linked, they tend to be linked in a loss collection manner such as software libraries. However, models all have submodels connected into whole by some logical structure; and these logical structures are theretically amenable to connecting into super-structures. In other words, in theory, any model could use the logic of some other models as their analytical components and be used by yet others in the same way. In actuality, of course, these models are seldom compatible in the way they represent their logical components, and hence can hardly inter-operate recursively ot be shared as common software resources for an enterprise or a community. We propose the concept of a Model-Base, in the spirit of the database model, to facilitate the management of models as enterprise or community resources. We also develop a particular structure to represent models into the Model-Base, using the Global Information Resources Dictionary method of the Metadatabase model. This work contributes to the lasting effort of integrating enterprise computing and information resources, and thereby making it easier to embed analytics in cyber-infrastructure for the benefit of the whole community.