RE: Is Semantic Web possible ?

Clearly the lack of usable ontologies is a serious weakness in the Semantic
Web project. To succeed, the Semantic Web will need a plentiful supply of
these datasets, and much more energy needs to be put into their development.
 
Early candidates were the philosophically "neutral" CYC and WordNet, which
avoid the obvious cultural biases of Dewey and the Soviet library, but
replace them with linguistic constructs that are really not very useful in
describing things. Classification schemes may be biased, but it may also be
more useful to know that information about stones is found under Geology
than in a section named "objects" (even for machines!). It has become clear
that no single large ontology can sensibly or comprehensively describe the
world. 
 
Equally, it is clear that developing ontologies by hand is difficult and
expensive. Look at the camera ontology that was reported to this list a few
months ago. That took time and effort, and it was very small. Look at
classification schemes like UNSPSC and the UDC with their groups of experts
and governing committees. And classification schemes are conceptually
simpler than ontologies. 
 
Unfortunately, software that generates ontologies automatically from text is
quite immature. It will improve over time, for sure. But the time scale may
be quite long. 
 
Who will ride to the rescue? Will Oxford University Press or Webster's turn
to developing the ontologies that give the Semantic Web its meaning? Might
Google, with its vast technical and data resources? Could Microsoft? Could
there be a new Open Ontology Project, uniting the ontologists of the world
.... ? Now there's a thought.
 
 

Bill 
.................................................................. 
Bill Hutchison, Chief Executive, Wordmap 
26 Upper Borough Walls, Bath BA1 1RH, UK 
T +44 (0)1225 358182 +44 (0)7977 454263 
www.wordmap.com 
..................................................................