And the back and forth continues...
>> Point us to these applications that can *currently* slurp in
>> 250 pages of full text and return 5 to 7 reasonably good, controlled
>> vocabulary subject headings (or topics or topic maps or well-formed RDF
>> triples or what have you).
> .. is a mish-mash of misconceptions and straw-man arguments. What
> exactly is "5 to 7 reasonably good, controlled vocabulary subject
> headings"? I'd like to know if a human could pass this test? Will
> there ever be a set of requirements of which this person will say
> 'yay' to automatic classification?
It's not a straw man argument at all. You have repeatedly assured us that AI can *already* do everything that catalogers can. This is what catalogers currently do. If they are to be replaced some time next month, I would like to know how.
As to why we would want the cataloging machine to spit out 5 to 7 headings -- my big concern as we look at changing the way we do description is interoperability. If we are attempting to index a big pile of records, it would be easier if all of them looked somewhat similar (as opposed to one pile having subject headings and the other pile having vector scores).
Again, nothing against technology or the knowledge domain of CS. But as T. Scott reminds us, AI has been just-around-the-corner for a long time now.
Received on Fri Aug 31 2007 - 09:29:00 EDT