Tuesday,
April 12, 2005
Global
Information Framework
Global Information Architecture, using
Composite Semantic Architecture
Prototype proposed by OntologyStream Inc
Draft Version 14.0 April 12, 2005
Principles:
Semantic Web technology falls
into two major schools of thought
a) First School of Semantic Science stipulates that ontology supports common sense reasoning with the imposition of constraint logics like OIL (Ontology Inference Layer for RDF – Resource Description Framework).
b) Second School of Semantic Science stipulates that ontology enables knowledge sharing, which can best occur with minimal dependency on constraint logics.
Simple Point of View: Ontology mediation of human knowledge sharing assumes no role for deduction or other types of logical inference. Ontology is simply a means to form community consensus and to make explicit some “sufficient” set of constructions.
Composite Semantic Architecture: OntologyStream Inc has convened a Science Advisory Board. The Board has made judgments about the complete set of primary semantic patents, and is capable of integrating the minimal configuration of patented processes and algorithms so as to provide an anticipatory technology based on total information exploitation in real time. Part of this minimal configuration involves digital signal processes. In this new technology a single step subsetting transform replaces data mining and pattern recognition. Consistent with the Second School of Semantic Science, this digital signal process involves no machine inference.
Provably Secure and Ultra Stable Operating System: The technical capability is demonstrated to have no scalability issues and is as fully agile as any RDF or XML repository. A complete architectural solution based on this work is available within a provably secure and ultra stable peer-to-peer knowledge operating system.
By visualizable we mean that a user interface will manipulate the conceptual constructions. The interface will operate in a fashion that parallels the human perceptual system, in the sense that a figure emerges from a background as an object of focus. Memory and anticipatory mechanisms are encoded as ontology. Clarity, consistency and completeness are promoted by this figure-ground dynamic.
By computable we mean that computer programs can manipulate the conceptual constructions. Envisioned manipulations include subsetting, search and to some lesser extent rules based on case or exceptions.
Purpose: This set of conceptual constructions aims to make visualizable and computable
· the aggregation of event information into a knowledge domain expressed as a set of concepts,
· the fetching of information using conceptual organization,
· the focusing of human selective attention using ontology subsetting mechanisms,
· the extraction of subject matter indicators from human text,
· the elements minimally needed for objectively examining risk and gain to the enterprise
The proposed architecture for Ontology Mediation of Information Flow is called Differential Ontology Framework ™ (DOF). DOF has the following elements:
1) A semantic extraction path that uses any of several COTS products to parse written human language and produce an n-gram (or generalized n-gram) over word stems, letter co-occurrences, or phrases; as well as rules defined over these n-grams.
2) A concept identification cycle that associates with the results of n-gram based semantic extraction one or more explicitly defined concepts.
3) An ontology development path that uses human descriptive enumeration to produce a three layer modular ontology – with the topmost layer highly abstract and common, the middle layer with multiple domain and utility ontologies, and a lower layer with small scoped ontology. (This is embodied in the FARES, Formal Analysis of Risk in Enterprise, product.)
4) Technical requirements are imposed on these three layers so as to, in the presence of instance data, produce a minimal subset of concepts from the middle layer that provides a clear, complete and consistent (the 3Cs) understanding of data reported as an instance and relating to an event.
5) Global aggregation of event structure so as to support global analysis of distributed and temporally separated events.