[protege-discussion] Ontology discovery

[Emanuel Warhaftig]

Hi, everyone,

In answer to Richard Donovan about Ontology Learning, I would recommend
looking at  the article "Ontology Construction for Information
Classification" by  Weng et al., in Expert Sytems with  Applications 31
(2006), for a novel ontology construction method based on Formal Concept
Analysis (FCA) (not a Machine Learning method, but a very interesting
method).

By the way, if anyone has additional interesting info. about the
connection between FCA on one hand,  and Automatic Construction
Methods of Ontologies on the other hand,  I would be happy to know about
it.

Thanks

Emanuel  Warhaftig

-----Original Message-----
From: protege-discussion-bounces at lists.stanford.edu
[mailto:protege-discussion-bounces at lists.stanford.edu] On Behalf Of
Vytautas Taujanskas
Sent: Friday, September 15, 2006 2:40 PM
To: protege-discussion at lists.Stanford.EDU
Subject: Re: [protege-discussion] Ontology discovery




Donovan, Richard <RDonovan <at> mtech.edu> writes:

> 
> I have been watching the Protege-discussion board for a good while now
(and
have not respoded before now)
> and have not seen any discussion in this regard.  I have however been
thinking
along these lines for some
> time and hve a firm notion that machine learning techniques can and
should be
deeply investigated for this
> application.  I have some experinece with ANNs and feel that they
would be
very applicable to developing an
> agent based application that could autonomously develop nominally
effective
ontolgies from
> unclassified data.

> 
> ________________________________
> 
> From: protege-discussion-bounce <at> SMI.Stanford.EDU on behalf of
Sione
> Sent: Fri 8/25/2006 5:22 PM
> To: protege-discussion <at> SMI.Stanford.EDU
> Subject: [protege-discussion] Ontology discovery
> 
> Hello,
> 
> Has anyone done any project or perhaps any publications about using
> machine learning or datamining to build an ontology for a set of
product
> that has not been categorised or say partially categorised?
> 
> Any hint would be appreciated.
> 
> Cheers,
> Sione.
>
------------------------------------------------------------------------
-
> To unsubscribe go to
http://protege.stanford.edu/community/subscribe.html
> 

There's a good research made by Ontology engineering group in
Universidad
Politécnica de Madrid. It's called "A survey of ontology learning
methods and
techniques" - you can get it there. 
(http://isk.ktu.lt:8081/wiki/?wakka=LiteraTure/files&get=deliverable_1.5
.pdf)

You need to define your data input and find appropriate method.

Vytautas Taujanskas


_______________________________________________
protege-discussion mailing list
protege-discussion at lists.stanford.edu
https://mailman.stanford.edu/mailman/listinfo/protege-discussion