Wednesday, April 12, 2006
Anticipatory Web discussionà [home]
Anticipatory Web Challenge Problem à [link]
Community Centric Service Methodology Glass Bead Games
Communication from Paul
Reto, and others
With this new discussion between myself (and my small group) and DERI principals, I feel that we have found the underneath of the layer cake (RDF) where perhaps some modifications/simplifications/corrections might occur.
We see the synthesis that DERI is making between OASIS and W3C standards. What seems to be happening is that the minimal complexity is being recognized and made clearly available as open standards.
I am including John Sowa and a few others into this discussion so as to outline some ideas. TBL may or may not be interested in this discussion, we have talked in person once. I communicate with a somewhat larger group in the bcc, and ask that if anyone wishes to post into this cc/to list, that the communication not focus on one's own viewpoint but rather on the general notion that a synthesis of OASIS and W3C might be possible.
Coretalk (Sandy Klausner) represents to me an ultimate model for the future web. Coretalk extends from Rosettanet, with a cubist design methodology. Ballard's work on information theory and n-ary representation of information structure also has certain appeal. But in both cases, there is a gap between this work and where the W3C and OASIS standards are.
I mention these two approaches simply for completeness.
First, having read Fielding's 2000 PhD thesis and understanding the quality of his advisors, it is easy to agree with the completeness of representational state transfer
http://www.ics.uci.edu/~fielding/pubs/dissertation/rest_arch_style.htm
as a basis for a provably secure, ultra stable (distributed) operating system.
The triple space is merely a set of RDF triples that maintain the basic subject verb predicate as a means to refer to resources. The two core problems that follows is the unique resource identifier (when "over" used "semantically") and the imposition of (as yet incomplete research on) description logics. I have stated, following the advice and literature of others, that both of these core problems are there due to an improper addressment of the Origin of Design issue. The Origin of Design issue is related to the Origin of Control issue and this issue to the problem of human induction of symbol systems (and "meaning").
Generalized from the RDF triple is a "syntagmatic unit"
< a, r, b >
used extensively in the Soviet era applied semiotics (Pospelov, Finn, Osopov) (see my on line book).. Generalized from the syntagmatic unit is an n-ary in the form
< r, a(1), a(2), . . . , a(n) >.
The nary is particularly nice in describing the clustering effects, or the categorization efforts of neural networks, genetic algorithms, latent semantic indexing, or Orb (Ontology referential base) transforms...
For measurement of co-occurrence patterns the n-ary is a nice output. And the n-ary can be used to automatically generate RDF triples. This transfer of information structure from a continuum technique such as neural networks or latent semantic indexing allows a distributed representation of information to be converted to a discrete representation of information (in fact in the form of a triple space). The imposition of description logic is problematic in most cases, but one can imagine a set of OWL Full templates where some nearest neighbor-matching algorithm between an output (small) triple space would select a template. Web service orchestration and discovery that starts with a textual input is the example I am thinking about and which is motivating the BCNGroup RoadMap developed for US Customs in late 2004.
Let me apologize if I am including anyone who does not wish to be included or if I am making some statement that is not correct when perceived from some perspective. I am often incorrect - but this work is not so easy to do.
Here is the core proposal. The "Web" is to be considered to have three axis, syntax, semantics, and pragmatics... following the literature in linguistics.
But, following the descriptions of de Moor, the pragmatic axis is closely tied to real time interpretation of an information structure by a human. Syntax is the structure of information, for example RDF syntax, n-ary syntax, natural language syntax. Semantics is meta data defined top down and not inclusive of pragmatics (which is emergent and bottom up).
Semantic Web, Syntactic Web, and Pragmatic Web can then be treated as distinct, and from which the union is to be considered the "Web".
The term "anticipatory" might be considered as an aspect of intelligence, something that can be operationalized if there is a measurement, encoding of invariance, and action cycle. But the issue of anticipation seems two steps beyond an agreement over what the "Web" is.
Reto (and Brahmananda) asked:
* what do you expect to be the interfaces of FSM?
* do you expect the interfaces to interlink triple graphs of different
namespaces?
FSM - Finite State Machine...
in two communication (which we edited together as:
http://www.secondschool.net/beads/communityCentric/10.htm
A technical discussion of these two questions will be developed and posted to
http://www.secondschool.net/beads/communityCentric/12.htm