(soon)
1/24/2004 11:52 AM
The next Semantic Web Interest Group (SWIG) meeting is scheduled for 1/29 at 3:00 at GSFC in Building 23, Room S-412. The guest speaker is Dr. Paul Prueitt. The title of his presentation is: "Towards the reality of a two-sided semantic web: the path to Differential and Formative Ontology".
If you plan on attending and do not have a badge which will allow you entry to
Goddard please let me know ASAP so that proper arrangements can be made.
I hope to see you at the SWIG on 1/29.
Walt
--------------------------------------------------------------------------------------------
Towards the reality of a two-sided semantic web: the path to Differential and
Formative Ontology
Dr. Paul S. Prueit
Founder: BCNGroup.org and OntologyStream.com
Presentation to Semantic Web Interest Group (SWIG)
NASA Goddard Space Flight Center
This presentation will cover some unique inventions that I have come upon over the past five years.
We all are aware of the gap between where we as a community think we should be and where we actually are. The brief description of a series of elegant and simple inventions will be helpful in mapping a path that goes at the heart of the unresolved issues related to standing up and using computer based representations about the natural world.
The first invention to be described is a scatter-gather algorithm on the circle, e.g. surface of a two dimensional sphere, that I invented in 1996 and published first in an Invited presentation in Moscow in 1997. This invention is surprising in that a one point compactification of the space leads to clustering that is not biased by boundaries. As in all scatter-gather methods, points are scattered randomly and then gathered using local information about relationships. The invention is exceedingly simple and yet has important generalization to scatter gather on the surface of a n+1 dimensional sphere if one have n degrees of freedom in ones theory of relationship. Software demonstration will be achieved using an in-memory mapping that is provably optimal in terms of performing millions of gather operations in a few minutes. Gruenwald discovered the invention used, but in a different context, sometime in the late 1990s and awarded a patent in 2003. This invention will be discussed briefly.
The next step is to introduce the notion of categoricalAbstraction (cA) and eventChemistry (eC) while demonstrating the software called SLIP, Shallow Link analysis, Iterated scatter-gather and Parcelation. I invented this work in 2001-2002. Don Mitchell, who developed the notion of a knowledge operating system, developed the code.
The next two steps are the introduction of the Subject Matter Indicator neighborhoods as a retrieval paradigm (patent pending 2003), and the generalization of the semantic primitive frameworks developed by John Sowa, Richard Ballard or John Zackman into a generalFramework theory (made public domain 2002). The last step is to discuss the notion of differential ontology and the notion of an ontology lens (invented in 2002)
The presentation will include a draft of a formal paper, (perhaps to be published elsewhere), a PowerPoint presentation, live software demonstration and free software.
Dr. Paul S. Prueitt
Research Professor
The George Washington University
Founder and Director, (1992) BCNGroup.org
Founder, (2000) OntologyStream Inc
Cell: 703-981-2676 paul@ontologystream.com
Paul Prueitt has established several commercial programs targeted at transferring and integrating data mining and text understanding technologies within a paradigm supporting knowledge management, emergent computing and decision support systems. He has an awareness of information technology implementation with the government and has provided consulting to agencies in regard to the declassification of 25 year old materials, the automated analysis of Islamic social discourse and the development of emergent ontological structure from the external analysis of document repositories. He has developed a virtual operating system based on number theory and used this to map the invariance of data structure into a linearly ordered in-memory map. He has a research background in knowledge engineering, neural networks, information extraction and retrieval, decision theory, data mining, distributed computing, and human factors design methodologies.
Dr. Prueitt is currently the President at Ontologystream Inc, where he directs work on adaptive technologies and knowledge management for industry clients. He is a Research Professor at The George Washington University, where he provides part time consulting directed at program development for the intelligence community. He serves as Director of the Behavioral Computational Neuroscience (BCN) Group, Inc. The BCN Group is a small, not for profit, scientific organization with a mission to aid scientific collaboration relating to machine and natural intelligence.
He discovered a general knowledge acquisition and management data base procedure called the voting procedure. The foundational work for this discovery is published and in the public domain. The voting procedure is based on situational logic and bi-level representation of object invariances in data sets. The voting procedure is consistent with a neuropsychological view of computational emergence and supports the formation of statistical inference and reinforcement learning.
Degrees:
· Post Doctoral Research
Artificial Intelligence, Georgetown, 1990-93