A semantic framework and software design to enable the transparent integration,reorganization and discovery of natural systems knowledge

I present a conceptualization that attempts to unify diverse representations of natural knowledge while providing a workable computational framework, based on current semantic web theory, for developing, communicating, and running integrated simulation models. The approach is based on a long-standing principle of scientific investigation: the separation of the ontological character of the object of study from the semantics of the observation context, the latter including location in space and time and other observation-related aspects. I will show how current Knowledge Representation theories coupled with the object-oriented paradigm allow an efficient integration through the abstract model of a domain, which relates to the idea of aspect in software engineering. This conceptualization allows us to factor out two fundamental causes of complexity and awkwardness in the representation of knowledge about natural system: (a) the distinction between data and models, both seen here as generic knowledge sources; (b) the multiplicity of states in data sources, handled through the hierarchical composition of independently defined domain objects, each accounting for all states in one well-known observational dimension. This simplification leaves modelers free to work with the bare conceptual bones of the problem, encapsulating complexities connected to data format, and scale. I will then describe the design of a software system that implements the approach, referring to explicit ontologies to unambiguously characterize the semantics of the objects of study, and allowing the independent definition of a global observation context that can be redefined as required. I will briefly discuss applications to multi-scale, multi-paradigm modeling, intelligent database design, and web-based collaboration.