[Sigia-l] Ontologies

[lisa colvin]

Hi Peter,

A formal ontology (ai/comp sci definition) consists of
a set of concepts ("standardized vocabulary") coupled
with formally defined relationships between those
concepts. The flexibility in defining the
relationships is contrained by the knowledge
representation language used to define it. 

My understanding of a controlled vocabulary is that it
is simply that: a vocabulary and not a framework for
defining formal relationships between concepts. 

How is this useful to IA? Good question. My work in
ontologies has been AI rather than IA :) The
ontologies that I've developed have been coupled with
an inference engine which can take advantage of
semantic constraints and inheritance. Clearly, natural
language processing systems need this kind of support.
However, are ontologies necessary or even useful for
typical IA projects? I suspect that for most cases,
not. 

Formal ontology development (with the intention of
using an inference engine) is very time-consuming as
much thought has to be put into the development of
relationships, constraints and their consequences. If
your project is for navigation within a website
(without need for inference), then a formal ontology
would be overkill. 

There is another discussion list on faceted
classification which I believe touches on the
usefulness of having a flexible
classification/relationship schema in IA-related
projects.  Perhaps some of those members have some
comments?

-lisa




--- Peter Cook <peter at clearscope.com.au> wrote:
> Would anyone like to comment on the relevance of
> 'ontologies' to IA?  I am
> practical applications to IA work.  What is the
> difference between an
> ontology and a controlled vocabulary?  Any pointers


__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com