- From: glenn mcdonald <glenn@furia.com>
- Date: Mon, 6 Jun 2011 10:37:13 -0400
- To: public-lod@w3.org
- Message-ID: <BANLkTin1RUWr=q2+YRtT_T_QA61L611E0w@mail.gmail.com>
It seems pretty clear to me that schema.org is a good step for data and humans. It's unlikely to be the end of anything, and I have my own set of particular issues and regrets** about it, but it's a potentially huge visibility/credibility boost for the ideas of structured data and common vocabularies. It's absolutely telling that it's already dominating discussion at SemTech, despite the conference being barely started here in SF. And maybe even more telling that it was done outside of the Semantic Web community. I think we (the SW "we") should find that embarrassing, not offensive. glenn **My top three reservations about schema.org in its initial form: 1. I *really* wish they'd given Thing an ID, separate from URL. I do not expect the URLs we get from this to be reliably usable as identifiers. 2. I wish they'd made types for everything, not mixed Things and "plain" values. But then, people do this in RDF all the time, too. I think it's one of the signs that the toolset still makes doing the right thing painful, just as people cram stuff into denormalized tables because RDBMSes make doing actual relational databases painful. 3. Honestly, I'm still totally not sold on the idea of embedding structured data in HTML. To scale, this stuff is going to have to be generated by the underlying data-management tools anyway, at which point I don't see why it isn't strictly better for everybody to have HTML for human consumption (unbloated by presentation-irrelevant data markup) and Data (in whatever serialization) for machine consumption (uncluttered by presentation markup/contingencies/layout). Even for Rich Snippets-like purposes this seems like it would be simpler to devise, produce and consume...
Received on Monday, 6 June 2011 14:38:01 UTC