Re: 301 Redirects to prevent some breakage in Linked Data

An interesting thought, Thad.

There are two aspects to this.

Firstly, there is navigation, by humans, around the web to the
documentation and pages describing the terms with said URLs.

Secondly, the URI identifiers for terms within a vocabulary which are
stored and programatically used to identify things.


In the first case I have some sympathy with your suggestion.

However in the second case, without specific inference rules built into a
system, no amount of 30x redirects will cause a lookup for the value of the
property http://schema.org/contentURL to return the value that was stored
for the property http://schema.org/contentUrl.

In Linked Data terms, they are different URIs.

Many of the major consumers of Schema.org currently blur the pure LD
conventions by inferring that http://schema.org/xxx URIs are the sameAs
https://schema.org/xxx URIs.  I have yet to see evidence of how widely this
assumption is adopted beyond the major consumers.  With a blanket inference
assumption such as http/https at least it is fairly simple to understand
and replicate.

With a more specific set of inferences, such as you propose, they are less
likely to be consistently understood and adopted across Schema.org
consumers in particular and Linked Data in general.

These bring me to conclude that applying such URI inferences could
introduce inconsistencies in the interpretation of Schema.org structured
data across the linked data web.   In that case, just introducing 301
redirects to the web representation of terms would lead to confusion as to
what is the canonical URI of a term.   Not only would not alert the lazy as
to their mistake, but also make it more difficult to identify such errors.

~Richard.



Richard Wallis
Founder, Data Liberate
http://dataliberate.com
Linkedin: http://www.linkedin.com/in/richardwallis
Twitter: @rjw



On Wed, 29 Apr 2020 at 16:29, Thad Guidry <thadguidry@gmail.com> wrote:

>
> *Problem: *"URL" versus "Url" is showing up in the wild west of our
> Linked Data web.
> Some properties, because of strict naming conventions we have, currently
> cause some breakage across the Linked Data web (but how much is anyone's
> guess).
>
> *Solution:*
> It seems there's a few Properties might be best served with a 301 Redirect
> ?
>
> https://schema.org/thumbnailURL  -->  https://schema.org/thumbnailUrl
> http://schema.org/contentURL  -->  http://schema.org/contentUrl
> http://schema.org/discussionURL  -->  http://schema.org/discussionUrl
> http://schema.org/downloadURL  -->  http://schema.org/downloadUrl
> http://schema.org/embedURL  -->  http://schema.org/embedUrl
> http://schema.org/installURL -->  http://schema.org/installUrl
> http://schema.org/isBasedOnURL  -->  http://schema.org/isBasedOnUrl
> http://schema.org/paymentURL  -->   http://schema.org/paymentUrl
> http://schema.org/replyToURL  -->   http://schema.org/replyToUrl
> http://schema.org/serviceURL  -->   http://schema.org/serviceUrl
> http://schema.org/targetURL -->   http://schema.org/targetUrl
> http://schema.org/trackingURL  -->   http://schema.org/trackingUrl
>
> *Questions:*
> 1. Could 301 Redirects be implemented for a few items on Schema.org?
>
> 2. Should 301 Redirects be implemented ?
>
>    - Maybe this makes developers and publishers too lazy?
>    - Maybe it helps consumers more, (things not breaking as often) and
>    that's good for all?
>    - How do others feel about that?
>
> Thad
> https://www.linkedin.com/in/thadguidry/
>