- From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Date: Tue, 06 Jul 2004 16:35:11 +0100
- To: SWBPD <public-swbp-wg@w3.org>, W3c I18n Group <w3c-i18n-ig@w3.org>
Note to I18N IG: 1) the Semantic Web Best Practices and Deployment Group works in public, please reply-all, but note that such replies will be public. 2) the [WNET] tag helps SWBPD participants to distinguish threads We are looking at mappings of the Wordnet theasuraus into OWL and/or RDFS. The basic approach is to map the conceptual relationships between words and senses etc used in wordnet (such as hypernym and synonym) into an OWL ontology, and to map a specific wordnet instance, such as the english one, into an RDF knowledge base acting as an instance of the ontology. In discussions in HP on this, we realised that we did not know whether a unicode string is a sufficient representation of a word, or whether XML markup is needed in some cases. The case presented in charmod is for using XML for *text* rather than single words. For example, ruby is not relevant to individual words, and as far as I can tell, bidi using XML markup is useful for text involving mixed languages and computer-ese etc. We were also trying to think about the issues involved with different languages making different uses of morphology versus the lexicon versus the grammar. A particular example that came to mind was long productive compound noun formation in german, where I assume that dictionaries do not list all known compound nouns ... (I am not a german speaker) Hoping for some I18N input ... Jeremy
Received on Tuesday, 6 July 2004 11:38:44 UTC