Tuesday, December 14, 2004

Waiting for the Schema Fairy

The Semantic WinFS "Instead, people will simply release metadata. Bad metdata. Good metdata. Incompatible metadata. Heck, they're already doing it without Microsoft's filesystem as a catalyst. Rather than a Schema Fairy, we need a standard method for generalized structuring of metadata coupled with a language for semantic translation. The good people of the world can release all the metadata they want, and when critical mass has been reached, somebody will create a translation between the two. If only we had some knight in shining armor to rescue us!

Have you ever heard those Trojan commericals where the Trojan Man rides up on his horse and saves the, um, date? Well, imagine that same commercial, except that the horse is the W3C, the Trojan Man is Sir Tim Berners-Lee, and instead of free prophylactics he's handing out RDF and OWL."


Via On the WinFS delay "If WinFS was only about figuring out how to plug in an object store to the file system, I think it would have a definite ship date by now. The technical issues involved in doing that are tractable, and the PDC build showed that MS is making decent progress on that front. The problem of associating blobs of metadata with blobs of information in the filestore is most likely done (or getting there). However, even with that in place there are larger issues: what does this metadata look like, and where does it come from?"

"Even if the Schema Fairy came down today and blessed us with canonical schemas for everything, there would still be the problem of metadata extraction and annotation. WinFS will only work if it has metadata to process, and right now there aren’t very many good ways of marking up a document with semantic metadata. Manual annotation just doesn’t work – the time cost to the user is too high."

No comments: