Dare Obasanjo's XML Metacrap - a small note
Dare Obasanjo's recent piece on Folksonomies, Taxonomies and Metacrap (shouldn't that be Volksonomies?) would provoke cognitive psychologists and computer scientists alike. Machine representation of semantics is preceded by knowledge representation (e.g., semantic networks) first and it defines the vocabulary also. This task has been ongoing from both sides of the fence - cognition and AI. Language semantics is an extremely difficult entity to emulate - ambiguity abounds (see Wittgenstein's language thinking concepts). Machinists have had limited success in a very limited capacity - Terry Winograd's SHRDLU in the "small blocks world" is a good example. I recall working on SHRDLU (in LISP) in a grad course in the mid-80s - it was quite amazing to see the machine respond to your queries (with meaning et al).
Current effort using XML tags are also represented in Topic Maps which shows tremendous potential but that too is restricted to the "small world knowledge domain". A fine example of domain expertise in HealthCare is SNOMED which makes healthcare knowledge and terminologies more palatable to the machines. Marry SNOMED with TopicMaps and you have some promise. But this marriage should be presided and blessed by the domain experts first rather than XML-Tag-happy constructs.
.