An Infrastructure for Information Discovery and Secure Knowledge Sharing

JohnRobert Gardner
Search, Browse, Taxonomy Tech Lead
Portal and Web Services, CSS-IT
Sun MicroSystems, Inc.

As portrayed in the film "Apollo 13," EECOM engineer John Aaron makes the point that, regardless the situation, that "Power is everything." Without it, nothing happens- it is paramount to the survival of the command module and the crew. Similarly, our research and work in the life sciences and high performance technology computing industries, concurrent with our commitment to open, loosely-coupled highly available architectures, has led to the development of a "deep knowledge infrastructure" approach which is to complex multi-industry integration projects (such as may be the case with the Virtual Iron Bird) what the neural sinews are to any living, evolving entity.

The advanced sciences and knowledge information technology engineers at sun have discovered a natural extension of the widely integratable Java Enterprise stack (Portal, Identity, Directory, Application Servers, and Search) in the area of dynamic knowledge management for the sciences. In these fields, just as in aerospace, the terrain is complex, uncharted or newly-discovered, and highly specialized in distributed vertical markets.

The proper integration of knowledge-based texonomy structures with discovery and concept mapping enables visual representations of what disparate systems are about with respect to each other, even when such a relationship is not actively known or input. This is where the combination of rich ontological structure and a subtle conceptual inferencing model (e.g., NOVA search in conjunction with swoRDFish) yield unprecendented introspection and association-revealing results. Our approach and architecture entails the mapping of multiple disparate taxonomies and capturing that mapping in an evolving knowledge base of rules which aggregates and, in turn, is applied on the fly to any and all new data. Expert review and identity management based access control and evaluation insure integrity and accountability. The process is iterative and the knowledge base derived is unprecedented in it's unlimited scope, scale, and integration with access control and contextual self understanding.

The implicit semantics in RDFS and RDF (domain, range, about, etc.) as a grammar for visual syntax in displaying the relationships between data nodes so marked has been known and proven in concept and application (Genomics) for some years. The integration of this capacity together with multiple mapped taxonomies in an inference framework which leverages portal frameworks for delivery and a rich identity (authentication/authorization) structure for managing collaboration and security is the "deep knowledge infrastructure" Sun has developed. We add to this greater value with leverage of the entire corpus of digitally-searchable hard-copy publication records.

Heavily leveraging namespace markup methods enables the knowledge base to integrate with visual represetnation in vector and 3D primitives so that discrete representations can "understand themselves" in context and relation to other components whose respective industries and ontological identities might otherwise be utterly unrelate. The patented algorithms of the NOVA engine can, in turn, be utilized to infer and develop a taxonomy from a corpus of specialized texts for industries where no salient ontology standard has been derived (e.g., aerospace).

Such a knowledge integration structure, more than any CAD representation, conceptual data, or visual navigational interface is heavily predicated and dependent upon its knowledge infrastructure. While inclusive of RDF, OWL, and related open and industry standards; it must also be augmented by a framework woven through all data nodes of all formats--graphic, data, numerical, text--in such a way that disparate systems and information not only know to what context they belong, but how their contexts relate to other contexts in an ongoing dynamic fashion to enable not just knowledge sharing and integration, but discovery of unanticipated connections, management of access, authentication and facilitation of collaboration.