The Semantic Web as the apotheosis of annotation, but what are its semantics?

Yorick Wilks

(University of Sheffield)

The paper discusses what kind of entity the proposed Semantic Web (SW) is, and
does so principally by reference to the relationship of natural language
structure to knowledge representation (KR). It argues that there are three
distinct views on the issue: first, that the SW is basically a renaming of the
traditional AI knowledge representation task, with all its problems and
challenges. If that is the case, as some believe, then there is no particular
reason to expect progress in any new form of presentation of this issue, as
all the traditional problems of logic and representation will reappear, ones
that have resisted definitive solutions within AI for fifty years. That route
will be no more successful outside the narrow scientific domains where KR
seems to work, even though the formal ontology movement has undoubtedly
brought benefits. The paper contains some discussion of the relationship of
current SW doctrine to representation issues covered by traditional AI, and
also discusses the issues of how far SW proposals are able to deal with
difficult semantic relationships in some well-defined areas of e-science.

Secondly, there is a view that the SW will be, at a minimum, the WorldWideWeb
(WWW) with its constituent documents annotated so as to yield their content,
or meaning structure, more directly. This view of the SW makes natural
language processing central as the procedural bridge from texts to KR, usually
via some form of automated Information Extraction. This view is discussed in
some detail and it is argued that this can also be seen as a way of justifying
the structures used as KR for the SW. There is a third view, possibly
Berners-Lee's own, that the SW is about trusted databases as the foundation of
a system of web processes and services, but it is argued that this ignores the
whole history of the web as a textual system, and gives no better guarantee of
agreed meanings for terms than the other two approaches. There is also a
fourth view, much harder to define and discuss, which is that if the SW just
keeps moving as an engineering development and is lucky (as the successful
scale-up of the WWW seems to have been luckier, or better designed, than many
cynics expected) then real problems will not arise. This view is a hunch and
not open to close analysis but one can only wish it well, as it were, without
being able to discuss it in detail further at this stage. In this paper we
will concentrate on the first three approaches, their problems and their
possible solutions.
Tuesday 19th February 2008, 14:00
Robert Recorde Room
Department of Computer Science