- From: Rushforth, Peter <Peter.Rushforth@NRCan-RNCan.gc.ca>
- Date: Mon, 8 Jul 2013 18:27:42 +0000
- To: "liam@w3.org" <liam@w3.org>, David Lee <David.Lee@marklogic.com>
- CC: mca <mca@amundsen.com>, "public-xmlhypermedia@w3.org" <public-xmlhypermedia@w3.org>
> Teaching Google's spiders that > <ingredient>honey</ingredient> is a link to > ../ingredients/honey/ is part of what I think is needed for > people to be able to use XML on the Web - right now if you > use anything other than (X)HTML with <a href="../honey" > class="ingredient">honey</a> you're doing business behind a > locked door. I've thought about this and come to a different conclusion. The Web allows resources to have multiple representations. Let the server inform the client what is available, through hypertext. Let the client decide which representation it wants, through content negotiation. The problem of google not understanding the mapping of <ingredient>honey</ingredient> to <a ...>honey</a> goes away, because I apply the ../ingredients/honey to <a ...>honey</a> mapping on the server, using XML technologies. Google and Bing's search crawlers do a fine job of indexing our html pages. Those pages have feed / entries backing them for machines and 'api' clients. If / when a crawler that understands XML (or JSON, for that matter) arises, we can serve them too. <a ...>honey</a> is text/html <ingredient>honey</ingredient> is not only application/xml but also most likely something like application/spl+xml. If a crawler 'understands' application/spl+xml, it will likely have an understanding of <ingredient>...</ingredient> and a mapping to a presentation structure will be unnecessary (for that crawler/search engine), because it would be built with that content and some presentation in mind. Cheers, Peter
Received on Monday, 8 July 2013 18:28:18 UTC