Friday, September 10, 2004
Background
material on why a National Project is required
Social Constructions and
Symbols à
Manhattan Project to Integrate Human-centric Information Production à
Hello Paul;
I read bead [107]. You have clarified the fundamental fact that there are no deductive chains in the operation of the Readware concept retrieval. We are relying on structure and theoretical conjecture about that structure.
I found your reply very clear and could not think of adding to it. Though I feel the need to defend my choice of words and I think I need to explain the phrase "wring the sense of any message or text".
In doing so, I hope I also clarify that we do expect (we know) that patterns cohere to their structure even when the names are changed. (My sense is that John Sowa’s work on similarity between graphic representations is focused on this issue precisely.)
To be clear, the Ant could be Martha and the Grasshopper could be a George-- the underlying patterns of the metaphor continues to exist as long as the network of relations continues to maintain coherence. If one substituted Martha for the Ant and a horse for the Grasshopper, for example, it is plain that some part of the network of coherence in the fable breaks-- if only because Martha and a horse are different species of entities.
The measures we propose could be used just like one would use destructive testing on steel to find out where it breaks. Such test can tell us where that internal coherence between phrases begins to breakdown. This approach to refining the substructural ontology for natural language is consistent both with Tom and my past work and with your suggestion that we use the qualitative structure activity relationship analysis (bead [75]) and quasi-axiomatic theory.
Some linguists are fond of pointing out how metaphors communicate ideas and I think that it was not well accomplished in this case-- if we take the term “wring” as the metaphor. Since I recently originated the phrase I can recall that I was thinking of a wet cloth, whereupon: while wringing out the water is possible, one cannot wring all of it out once and for all. In like fashion, we do not wring all the sense out of messages or texts. As you clearly stated we are only providing a framework, capability and infrastructure for doing that 'sense analysis'.
In fact, in my originating message, the phrase was "...a universal analytical infrastructure upon which they can wring the sense out of any text or message and compare any to any as if all messages...".
To "wring out the sense," we do now already list out the words, names, numbers, idioms, important phrases.
We link them to known or reference concepts, topics and subjects of investigation and of judgment, we classify instances and occurrences of known patterns (partial phrases, noun phrases, disambiguated names/nouns (Turkey=nation)) , and compile all this into a compact mathematical signature to be indexed and used for further analysis.
I look forward to Bill’s comment and to John Sowa’s comment.