Re: Post Type Discovery

On 10/07/2015 09:19 AM, Christopher Allan Webber wrote:
> elf Pavlik writes:
>
>> I added issue on IWC wiki
>> * https://indiewebcamp.com/post-type-discovery#Issues
>>
>> I see current claim of compatibility with AS2.0 very misleading and
>> purely based on included in AS2.0 drafts examples of Microformats HTML
>> serializations, about which James wrote a clear NOTE in both specs:
>> "The Microdata, RDFa and Microformats examples included in this document
>> are purely informative and may not currently reflect actual
>> implementation experience or accepted best practices for each format.
>> These alternate serializations may be removed from future iterations of
>> this document and moved to a separate informative WG Note."
>>
>> As of today Post-Type-Discovery only applies to modeling used by
>> participants of IndieWebCamp and assumes use of Microformats Vocabulary.
>>
>> I don't say that I support modeling used by IWC based on Microformas
>> Vocabulary or that I support modeling used by James based on
>> ActivityStreams 2.0 Vocabulary. I just think that pretending that those
>> two mentioned use compatible models, and that proposed 'type discovery'
>> supports both, only brings more confusion to current state of things.
>>
>> If supporters of this draft really want it to support both Microformat
>> and ActivityStreams 2.0 based modeling, I see appropriate to show it
>> with examples which use *both* recommended AS2.0 modeling and
>> recommended Microformats modeling. As I see it this will require
>> modifying at least one or both models.
>> https://indiewebcamp.com/post-type-discovery#Examples
> Thanks, elf!
>
> What if the document is clarified clearly as "infering types for
> Microformats or domains lacking type information"?  If we can clarify
> that scope, I think it might be useful in providing ways to help
> Microformats people move their stuff to ActivityStreams, which could be
> useful for reducing friction in the group.

Note that the microformats community, as well as the JSON community at
large, does *not* use RDF inference. I assume also by RDF inference you
mean RDF(S) inference, as there's a plethora of largely incompatible OWL
inference schemes.

In my opinion, from my experience with RDF(S) inference a while back at
Yahoo!, its very hard to get to scale with big data and doesn't work
well on messy data - and that's not even OWL data. I think it's safe
also to assume RDF inferencing will get better and RDF tooling in
general. However, until then - there's probably a role to play for
making sure people know precisely what to infer, both for JSON users who
don't have RDF tooling and the microformat community.

What we are trying to do is to bridge the gaps between communities. So,
as long as the results of the type discovery (inference) procedure are
the same, it's useful!

  yours,
      harry

> I do think it's true that once you already have a linked data /
> activitystreams type representation, you've probably already set the
> type, so you probably don't need this document.  But if we can reduce
> the friction between microformats->activitystreams that would be helpful
> for the group, I think, and might help reduce some common arguments
> while giving us a clear path towards deliverables, which I would really
> like to see.
>
>

Received on Wednesday, 7 October 2015 13:43:49 UTC