First, a few core principles.
The nature of human, or animal, intelligence is far more intricate and complex than what RDF (Resource Description Framework) and most Semantic Web standards assume.
The current practices in data storage, analysis and management are crude and far more cumbersome than any one has a right to imagine. All data encoding can be translated into a set of syntagmatic units each having the form < a, r, b > where a and b are “locations” and r is a relational operator. If this is done well, most databases can be written in such a way that information is more quickly recovered and memory requirements are reduced. The set of syntagmatic triples
{ < a, r, b > }
are schema independent, and yet an inverse translation process is always available to re-recreate any relational database schema and re-populate that schema with the original data. Data cleaning and referential integrity can be achieved in the data schema independent state, so that when the relational database is reconstituted the data will have higher value. Specialized data stores for graphical rendering can be constructed. Any two such databases are easily merged, and selective reading of a single database can produce a splitting of the data.
Data processing might optimally be achieved most efficiently using two simple innovations. The first is a key-less hash table were the “hash” function is simply interpreting an ASCII string as if it where a base 64 number. The second innovation is the separation of a measurement of structure, via co-occurrence and other structural relationships; from an interpretation of the situated meaning. This second innovation does not discount the value of having lite ontology and taxonomy to automate disambiguation and focus on subject indicators. The notational paper on Orbs and the notational paper on SLIP present technical details. Both innovations are operational in software. Zero dependencies exist on third party software. We found that using a Berkeley Open Source hash table management system for design was useful. Switching from the standard hash table to the key-less hash table is a simple substitution.
National needs are to be properly addressed by looking at how data processing might optimally be achieved most efficiently. We are asking In-Q-Tel to make a small start-up investment in OntologyStream Inc so that a major advance in computer science can be demonstrated in the context of aiding human knowledge propagation within secure collaborative environments and in providing a new capability called Human-centric Information Production (HIP) intelligence.
Intellectual property has focused on less than optimal solutions because an optimal solution most often takes the form of very simple mathematical constructions. Even in 2004, pure mathematics is not easily patented. The OntologyStream proposal is to make simplification to all relevant patents so that the underlying foundational concepts are expressed as Open Source information and through Open Source software. Once these concepts are expressed they can be brought together into a virtual collaborative environment based on the concept of the knowledge sharing foundation. Successful business models will depend on the details of an application, such as in pharmaceutical research and production and in specific analytic work on intelligence. As such, the business models shift from control of the data by vendors to the provision of expertise in new manufacturing and distribution systems.
This shift is not anticipated by Industry. Significant effort is needed to help prove new business models will be successful when the work being done is not in the development of data processing infrastructure, but rather in the manufacturing or distribution of product. The positive aspects of knowledge management literatures has established some of the preliminary principles, such as how lessons learned are to be shared and how terminological differences might be subjected to reconciliation processes.
The ultimate objective of my work is to help create a process producing a Knowledge Science curriculum for K-12 and college.
Inference and workflow are to be achieved using a stratified model based on my understanding of a discipline called applied Russian semiotics and on a mathematical-logical formulation called quasi-axiomatic theory; and on my understanding of cognitive neuroscience, theoretical immunology and stratification theory. The work on theoretical immunology was extended form one half of my PhD thesis as a model of self / not-self complexes developed by children with respect to middle school and high school mathematics curriculum. A model on learned learning disabilities is suggested.
Application areas are numerous. However, these areas now struggle with cumbersome databases and information theory. The promise of SLIP and Orb technology includes foundational work for global weather-map type event detection and display in areas related to disease outbreak and spread, tracking computer virus infections, and mapping the thematic structure of real time social discourse (from blogs, web news sites, chat rooms, or email traffic.) SLIP and Orb technology is easier to understand, easier to use and is mostly un-encumbered by proprietary intellectual property. (A few problems exists here.) Given one major example of Orb and SLIP deployment, a great number of new applications of the foundational concepts will not be inhibited by extensive proprietary constraints on the basic capabilities. At that point, the justification for the National Project to create a Knowledge Science curriculum will be evident.
The key application will be to the aggregation of blog themes into a large-scale real time conceptual index. The index will have individual portals that behavior to record analysis and preference patterns. The thematic analysis of these patterns will produce anticipatory web services, for the first time.
Using sets of syntagmatic units
{ < a, r, b > }
written in the key-less hash table, data fusion exists in analog to how Karl Pribram’s theory of coherent electromagnetic interaction, mediated by neural control, suggests cognition occurs. Distributed, virtual Human-centric Information Production (HIP) is provided a simple, and completely transparent foundation. There is no business model, just mathematics.
Operational software is available for review