[26]                               home                            [28]

ORB Visualization

(soon)

3/2/2004 7:20 AM

National Knowledge Project

 

difference between CCM and Orbs

 

 

I woke up in the middle of the night thinking about the difference between CCM and Orbs.

 

I have been so intensely working on the Orb concept that I have lost some of the awareness of the details of the CCM patents.  However, it would seems to me (although I could change my mind if given argument that was indicating otherwise) that the CCM patent, when applied to text analysis, absolutely depends on having these short branches between the center of the word level n-gram and the first significant word to the left and the second significant word to the left - and likewise to the right.

 

http://www.bcngroup.org/area2/KSF/Notation/notation.htm#_Section_2.1:_

 

The Orb, however, has a fundamentally different construction.  The center word is regarded as the center of a topological neighborhood with all of the elements in the neighborhood having the same "nearness status.  So there is no branch construct.

 

Once this change is made things are very much simpler, both in the notational description and in the encoding of data.

 

One only needs to form the category of all occurrences of each word while preserving the relationships between centers and words in the neighborhood.  I was doing this in 2001 with the SLIP browsers and thus was able to write a file with the triples

 

{ < a, r, b > }

 

where a is a center and b is an element in the neighborhood.  The file is produced from the n-grams or generalized n-grams.  The generalized n-grams work with the generalFramework theory and Zackman type frameworks - as I am beginning to discover. 

 

The introductory work on frameworks is at

 

http://www.ontologystream.com/beads/enumeration/gFfoundations.htm

 

and the use of a generalized n-gram is natural.

 

The SLIP browsers automatically produce the graph constructions that I called an Orb (Ontology referential base).  Nathan Einwechter coded the construction process using a simple hash table management system.  The SLIP browsers actually have the keyless hash table (I-RIB) that generalizes the Primentia patent, and so we had all of the pieces of this very powerful technology.  My sense is that the generalizations of CCM and Primentia patents are actually not infringements and are actually more powerful. 

 

What I did not have is a way to market, or a business model. 

 

The use of set theory and mathematical topology may seem exceedingly abstract, but it is very natural to someone who has studied mathematical topology (which is in some special way a formalism on the notion of nearness.)  And the construction is intuitive and easy to explain. 

 

What I suspect is that because the most basic construction is a set rather than a graph tree branch, the Orbs do not infringe the CCM patents.

 

I am saying this speaking as a mathematician and only with the formalism in mind.  My sense is that the best way to deal with my own inability to develop a business model around this is to make the next construction public domain - which I already have to a large extent (although I have one year after publishing to file a patent or provisional patent.)  This next construction links to RDF and to inferential structure regarding the placement of things into categories (an semi-automated ambiguation/disambiguation method.)

 

The differentiation of nearness into types of nearness is utterly unexplored territory. 

 

The current state of I-RIBs and Orbs are enough already to really excite people that there is something to human information production using schema independent data encoding.

 

The work that I did for the FCC was really important, as it tied together one a most reasonable use of Orbs as a means to generate and deploy a topology of subject matter indicators:

 

http://www.ontologystream.com/beads/InORB/home.htm

 

I knew that the FCC IT managers would never understand Orbs so I brought to them a demonstration of the Stratify technology and polling methodology until they saw what to do, and then they decided that the project really did not need to reveal the internal deliberative process to the degree that actual indexed taxonomy would do, and so the project ended.