5/21/2004 7:27
PM
On the generality of the particular
Key questions on Common Upper
Ontology -> .
Amnon,
Your TAIParse is an application for producing text analyzers. The importance of your patent application has always been apparent to me, ever since my days as senior scientist at one of the intel consulting companies. In fact the use of TAIParse application for producing applications fits well within the nine-step actionable intelligence diagram that I developed from a seven-step slide. The seven-step slide represented the understanding that many in the intell consulting community had, and continues to have, regarding the steps needed in any community-based production of actionable intelligence.
The seven-step slide missed the first two steps: {instrumentation/measurement} and {encoding/interpretation} of the nine-step slide. I have made a big deal about this in several forums. In fact, the very concept of an Anticipatory Web of information demands real time management by a human or human community over things like thesaurus, controlled vocabularies, and ontologies.
My voting procedure can be applied to guessing membership in categories due to word occurrences. Once one begins to use this type of process, one finds both a simplification of the technology for text unit routing into categories and a few nice surprises related to inferencing about the second category given the user (or some decision mechanism) glances at the results and "knows" that the first choice is not proper.
The Human-centric Information Production (HIP) aspect of the Anticipatory Web technology is again made apparent in the Prueitt voting procedure.
I developed the voting procedure in 1996, but it has never been deployed - in spite of the clarity of the public descriptions. I did operationalize the system in 1996 (automated declassification design technology) and then again in 1999 (distance learning design technology). The problem that we have been working on is about how to communicate the essentials of Anticipatory Web concepts.
WorldFuture 2004 is held this year in Washington DC, July 31-August 2. At WorldFuture I will give a presentation. I am thinking about writing a short booklet and have in sold during WorldFuture 2004. The Anticipatory Web concept has come to be one that is being discussed as an alternative to AI and Semantic Web. I would like to make the booklet the definitive source on Anticipatory Web principles.
Your master plan fits exactly with how several of us see things should be done. Of course, there is no way that I personally could develop even a single (good) text parser by hand, much less an application that assists in rapid development of situational text parsing.
As we have discussed in the past the parsing does not need to be over ASCII text. This opens the door to ontology parsing, whether the ontology is Semantic Web OWL or Anticipatory Web Orb format.
A small group continues to dream about what might be done with your work and five or seven others. As you know, the knowledge sharing core concept might rapidly integrate advanced and un-encumbered capabilities into a single system. This single system might then be used by average people in the development of knowledge about how computer text analysis systems work and what can be done with such systems.
We do sometimes get the attention of folks who understand the pragmatic axis related to real time situational analysis. It is clear to them that the critical need is for a computational separation of
1) a theory over categories on invariances in the past event measurements, and
2) the very different anticipatory frames that reflect environmental conditions as well as the long term plans by individuals and organizations.
But this need to "build capabilities for a particular task and domain" (as you say) is almost completely ignored by those who run programs. (It seems to me.) And thus there has been up until now positive peer review but no funding.
Some who think themselves important in the Semantic Web community cannot understand this separation, or the grounding that the separation has in natural systems.
But that could change in a heartbeat. Our problem is really one of educational background. The National Project will address this educational issue as the first task.
In response to your comment:
I
know you have methods for dealing with small documents sets and the like,
thought you might have some feedback on all this. At some point, a few months down the line, I'll want to show you
what I have
Yes, small document sets are quite different from large document sets. One needs both types of technology.
As I have said many times, if one only has technology that is stochastic or statistical, then one has a powerful and interesting tool. But this technology can misinform our analysis simply because we, as a general rule, do not fully take into account what needs to be taken into account.
A true comprehension is needed about the nature of the relationships that the past and the anticipation of the future have with the present moment experience. This comprehension is still in the distant future.
The pragmatic axis can only be seen in the particular and the particular only exist in the present moment.
The bead games continue to be developed:
PSP