W3C home > Mailing lists > Public > semantic-web@w3.org > October 2010

Re: rdfa vs. links

From: Nathan <nathan@webr3.org>
Date: Sun, 24 Oct 2010 23:43:43 +0100
Message-ID: <4CC4B69F.3000005@webr3.org>
To: Lin Clark <lin.w.clark@gmail.com>
CC: Stephen Williams <sdw@lig.net>, William Waites <ww@styx.org>, semantic-web@w3.org
Lin Clark wrote:
>> With RDFa, when a user cuts and pastes visible HTML content, they also get
>> the RDFa that is exactly associated with that content.  There is a demo of a
>> Javascript page that can receive the paste and display the RDFa nicely.
> I'm a proponent of RDFa, but I actually see this particular behavior as a
> bug, not a feature. Because the RDFa is hidden, you can easily copy text
> from the Web and paste it somewhere where the hidden tags will make
> incorrect assertions.
> For instance, when I copy and paste a co-worker's name into a page on my Web
> site, it would copy the foaf:name property. The foaf:name property worked
> well on my coworkers Web site, where the foaf:name took the URI defined in
> the parent element as it's subject. However, when I place it in an arbitrary
> position on my page, it will then take another element for it's subject...
> for instance, it might be pasted into a div about me, in which case it would
> assert that my coworker's name is also my name.
> I'd be interested to hear what other's think about this.

Personally I see this more as a "currently poorly supported feature" 
rather than a bug tbh.

For me RDFa opens the door to a world of new applications, modular 
applications which understand parts of the data on a page and can 
optionally augment the users view with more information, or allow them 
to do something with that information, use-cases range from simple 
things like "your friends x & y know the author of this post" and "good 
relations powered clients side shopping basket/wishlist" (which would 
work over any website) through to more complex applications like 
automatically aggregating your thought streams based on what you're 
reading/viewing (you are what you eat) providing you with an audit trail 
so you can see exactly where your good idea(s) came from, and where 
they're headed.

I can't see any black or white best practices here, in one use-case it's 
good to have lightweight machine readable data (ala turtle/n3 when only 
a machine is considering the data) in other cases it's good to have 
hybrid human and machine readable data, especially in an agent augmented 


Received on Sunday, 24 October 2010 22:44:25 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 08:45:20 UTC