Monday, April 25, 2005
On the discussion of data
regularity à
On the Stephenson Cyber Attack
Taxonomy à
Alex sent the
following communication
Paul,
Our yesterday's conversation gave me food for thinking. I feel that I
understood many things, but I am still not clear in certain cases. Could you,
please, explain to me the following:
Let's assume that Customs officers suppose to fill in the report based on the
following template:
Date of event
Point of Origination
Destination
Consolidator
Shipper
Consignee
Reason for checking
Findings
Officer's Name
Officer's Signature
In this particular case what kind of stems would be used (examples), what kind
of 7-tuples can be produced (examples), and what kind of analysis can be
performed (example).
Thank you.
Alex.
Alex:
There seems to be something missing in your questions here.
As before, the problem is that I have not stated clearly the ontological assumptions that Adi makes in his Part I - III
A set of PowerPoints are available on the Custom’s work.
We, Adi and I, assume that a theory of the world is available through a theory of how the Arabic language works.
Once one has this theory of the world, one is able to develop theories of how each of the other languages work.
Most working linguists will have problems with this assumption, but not all will. There is a history here that is full of controversy. I make the argument (below) that the ontological assumption that a theory of natural language can be found is not completely necessary to the use of semantic extraction processes defined in NdCore, using probabilistic Latent Semantic Analysis (form Recommind’s patents), or from Readware (as Adi reveals in Part I – III).
One can use pure observation – if one has a type of representational language to preserve the observations made. This is how chemistry was developed, and also as with chemistry we need an ontological assumption that the natural processes in the real world stratify into organizational layers.
The methods I have defined in differential ontology framework, DOF, would certainly help exploit this approach. The DOF approach is empirically based (one has to talk with Tom for a while to see how he did actually, over many years, experimentally assigned ontological concepts (with ambiguation/disambiguation metadata along with the assignment) to about 3,500 Arabic stems).
This experimental work is done not with fixed rules but as a chemist would do early in the development of chemistry (ie through experimentation and observations). Over time some weak rules where discovered, and then formalized into a type of string theory for Arabic. Underlying this string theory for Arabic, we have a theory of the natural social world. This is a theory that has not as yet been shown to be “correct” through a careful process in which the empirical work is made objective through some sort of formalism.
Adi’s work was completed without DOF, but his work followed the intent that is in DOF.
Using DOF, the empirical work from semantic extraction processes is transferred to a formal encoding of
1) explicit description of concepts (using words and relationships to other concepts)
2) metadata related to ambiguation/disambiguation issues related to “that” concept
3) cultural use, reconciliation, issues related to “that” concept
This formal encoding can be done using standard OWL lite, or simple RDF or simple Orb constructions, see roadmap.
RDF is XML having the form < object, verb, predicate>. Orbs are more general and simply refer to the abstraction notion of two graph nodes and a relationship operator: < a, r, b >.
Orbs were first developed in the former Soviet Union, based on the work of Pospelov and others. They are very simple in nature and allow a definition of a theory of convolution operators that work in computer memory structures as structural holonomy.
It may have been possible to have originally stated this ontological assumption about the nature of language with a language other than Arabic, Chinese for example. Perhaps this type of statement already exists in certain works by Chinese linguists? Perhaps “semantic science” is more developed elsewhere.
Ok now to your excellent
question.
If some of the fields are full text, then the normal process of creating sentences will use only certain types of words - reflecting underlying letter use semantics.
In Arabic this development of first a stem and then filling with vowels is reinforced by culture, but a transcription of the underlying theory of the world, the social world, is developed in Readware using what they call “cultures”. The cultures are specific interpretations about classes of English (all consonant) stems, like “wrk” for work and working.
Adi's stems are composed only of consonants, and mostly only having three letters. See the Q matrix. in Part 1.
It is a bit esoteric to suggest, as Adi and I do, that something about the use of consonants is universal. I make the argument than Arabic cultural use is merely a good manifestation of something that is available in the physics that human beings and the world are built up from. So the tie in to my work on Pribram’s theory of perception and the holonomic theory of brain function is available to us.
Adi presents his work in an authoritarian fashion, as he should and as necessary. However, he and I have spent hundreds of hours talking and we both know that this theory of language is very early in development.
On funding issues:
I would like some help in re-framing the road map into a response to
http://www.darpa.mil/ipto/Solicitations/open/05-28_PIP.htm
Hopefully we will get some support in this proposal writing from NMSU linguistic consortium (Rex Brooks, will you forward this note to those you know). The proposal is due May 3rd).
Funding is essential. I am looking for some consulting on event analysis and predictive analysis methodology using DOF. Many in “our” community are under-employed or not employed at all. The great need for event analysis and predictive analysis methodology has been rather completely filled with business processes that use the need but offer less than the solution that we see is “easily” available to the government, and to society.