[WNET],[OEP] OntoWordNet. A new large OWL ontology

Dear WNETters,

I've finally managed to upload the promised 
material to the [WNET] page: 
http://www.w3.org/2001/sw/BestPractices/WNET/. 
This is the first of a series of messages 
describing the new material.

This message deals with OntoWordNet, an OWL model 
that reengineers the noun synsets from WordNet 
1.6 as a formal ontology, aligned to the DOLCE 
foundational ontology. It's downloadable from: 
http://www.w3.org/2001/sw/BestPractices/WNET/wnNounsyb_v7.owl.

Best
Aldo

_________________
This is a preliminary version (v0.7) of the OWL 
alignment of WordNet 1.6 Noun Synsets to the 
DOLCE-Lite-Plus ontology library.

For details on the DOLCE-Lite-Plus library, see 
http://www.loa-cnr.it/DOLCE.html.
If you want to browse DOLCE-Lite-Plus, and 
looking at the axioms, download the latest 
version (3941), from the DOLCE site (but I've put 
it in the [WNET] page as well: 
http://www.w3.org/2001/sw/BestPractices/WNET/DLP3941.owl, 
and either load it on an OWL editor (to browse 
separately), or import it into OntoWordNet (to 
see the alignment with axioms, comments, etc.).

This file takes about 3 minutes to load into 
OilEd, and 8 minutes to load into Protégé (both 
measures taken on a PowerBook G4 1.5 GHz, 1Gb 
RAM, allowing 630Mb of memory to both OilEd and 
Protégé). On the other hand, OilEd does not seem 
able to visualize the entire taxonomy. In 
Protégé, saving modifications requires some time.

This OWL version, in order to keep an acceptable 
dimension, only contains taxonomical information 
and the original glosses (as comments).

The alignment is part of the OntoWordNet programme, which aims at:

1) reengineering WordNet lexicon as a formal ontology, and in particular:
	1.1) distinguishing synsets that can be 
formalized as classes from those
		that can be formalized as individuals
	1.2) interpreting the hyperonymy relation of WordNet as either
		owl:subClassOf or rdf:type
	1.3) interpreting lexical relations from 
WordNet as ontological relations
2) aligning the top-level of WordNet to a foundational ontology, allowing for
	re-interpretation of hyperonymy (as owl:subClassOf) when it is the case
3) checking the consistency of the overall result, and correcting the cases for
	inconsistency
4) modularizing the resulting ontology into modules according to an initial
	proposal of 'domains', which will be 
revised and reinterpreted according to
	logical and formal ontological principles
5) learning and revising formal domain relations 
(either from glosses or from corpora)

Until now, the tasks 1.1, 1.2, and 2 have been 
fulfilled as far as Noun Synsets are concerned.
More than 60,000 OWL classes, and almost 5,000 
OWL individuals have been created. About 20 
classes and 10 individuals still need corrections 
and will be included in the next version.
Almost 1,000 classes have been aligned to 90 
DOLCE-Lite-Plus classes, in order to make sense 
of the main peculiarities of WordNet taxonomies 
when formal axioms constrain the intended meaning 
(as derived from DOLCE-Lite-Plus).
Such alignment has been performed by analyzing 
the WordNet 1.6 noun synset taxonomy until the 
third or fourth level. In many branchings, lower 
levels are semantically consistent, while in 
others, I expect that further refinement is 
needed.

The task 3) has been partly done, but corrections 
are not yet in place in this version (about 700 
noun synsets of WordNet 1.6 have multiple 
hyperonyms, and some inconsistencies after 
alignment to DOLCE-Lite-Plus have been 
discovered).

Ongoing work is on tasks 1.3), 3), 4), and 5), 
and on the reengineering and alignment of the 
Verb and Adjective Synsets. Moreover, a mapping 
of OWN to WordNet 2.0 is envisaged in the short 
term.

For some of the principles underlying the 
alignment work and the OWN programme, see:
- Gangemi A, Guarino N, Masolo C, Oltramari A, 
Restructuring WordNet's Top-Level, AI Magazine, 
Fall 2003.
- Gangemi A, Navigli R, Velardi P, The 
OntoWordNet Project: extension and axiomatization 
of conceptual relations in WordNet, Meersman R, 
et al. (eds.), Proceedings of ODBASE03 
Conference, Springer, 2003.

The work is indirectly funded by academic and 
industrial projects, but the programme has no 
specific funding. For researchers interested to 
collaborate, please contact Aldo Gangemi at 
a.gangemi@istc.cnr.it.
-- 



Aldo Gangemi
Research Scientist
Laboratory for Applied Ontology
Institute for Cognitive Sciences and Technology
National Research Council (ISTC-CNR)
Via Nomentana 56, 00161, Roma, Italy
Tel: +390644161535
Fax: +390644161513
a.gangemi@istc.cnr.it

*******************
!!! please don't use the old gangemi@ip.rm.cnr.it
address, because it is under spam attack