Another important area that has been given
considerable attention in recent years is to explore
how existing KOS could be employed and made
us of in the web environment, Faceted navigation
has been recognised as an effective way forward
by information architects and web designers. There
have also been efforts at porting KOS to Web
(Semantic Web?). The SKOS has been designed to
provide a low cost migration path. It also provides
a conceptual modeling language for developing and
sharing new KOS. It can be used on its own, or
in combination with more formal languages like the
web ontology language (OWL). The role of SKOS is
to bring the worlds of library classification and web
technology together. Automatic classification based
on supervised methods appears to perform better.
However, availability of large training corpora for
different domains, different languages, and different
classificatory tasks is a major issue. For example,
in the area of WSD (Word Sense Disambiguation),
it has been estimated that to achieve reasonable
levels of accuracy one needs about 3.2 million sensetagged
words. The human effort for constructing
such a training corpus has been estimated to be
27 man-years! A significant portion of the work is
in experimental stage and several approaches are
being experimented with. When tools and applications
are designed to work focused, on a specific domain
of interest, the results appear to be better. Figure
1 presents an overview of the various approaches
to text classification vis-à-vis the cost and degree
of automation that can be achieved