Wednesday, September 01, 2004
Manhattan Project to Integrate Human-centric Information Production
Stratified complexity and the origin of mental/social events, (Prueitt 2002)
A discussion between information scientists
The notion of a "periodic table" underlying human concepts expressed in language ties in with the notion of "semantic primes and universals" as elaborated by many thinkers, e.g. by Anna Wierzbicka who identified 20 or so primitive concepts and tries to express all possible concepts using these primitive concepts plus information about sense perceptions. I think this is a reasonable approach, but that the atoms will never be as unique or clearly-defined as in the chemistry case: I think you can isolate many different "conceptual periodic tables" underlying human concepts as expressed in language.
I tend to doubt that statistical text analysis of the LSI kind will be able to isolate semantic universals -- but I do think that statistical analysis of conceptual structures extracted from texts using language parsing followed by automated inference, might be able to do the trick. So I think your idea of extracting these semantic universals through large-scale automated or semiautomated text analysis may be sensible, even though I'm not confident in your choice of technology.