On 2/6/12 10:34 AM, Huwig,Steve wrote:
> While it's generally more helpful to have readable URIs than to have
> machine-generated ones, I think that effort expended to standardize URI
> construction won't be very effective. Naming and organizational
> conventions in one culture or system will likely not translate well to
> different cultures or systems.
I actually wasn't thinking about readable v. opaque URIs, but that is
one topic for best practices. In my experience, you get loud complaints
with opaque URIs, but readable URIs are highly problematic.
I was thinking about things like:
- "cool" URIs - hash? or no?
- PURLs v. agency domains
- what should the URI resolve to?
That kind of thing.
> In my opinion, the most intellectual effort in this area should go into
> defining a bibliographic record media type, or a family of media types,
> or a framework for creating and using such media types. The idea is that
> if the automated systems have unambiguous ways to process these media
> types, then the users can expend as much or as little effort as they
> need on naming and organization.
> Naturally, the media type will use URIs and hyperlinking prolifically,
> but in the end computers don't really care what the URIs look like as
> long as they can be resolved. In fact, specifying URIs to be opaque in
> the data model will be very helpful, as then we can be sure that the
> transmission formats and system implementations won't need to rely on
> potentially ambiguous or contentious naming and organization schemes.
> More on URI opacity: http://www.w3.org/TR/webarch/#uri-opacity
> Steven Huwig
[log in to unmask] http://kcoyle.net