Re: dwbp-ISSUE-79 (metatype): Discovery vs structural metadata [Best practices document(s)]

Hi Makx,

I agree with you that's gonna be hard to reach consensus about  fundamental
metadata questions, like "what metadata is, what kinds of metadata are
relevant, and what metadata should be provided in an ideal world".

I really like the idea of providing best practices for metadata on a
highest level instead of trying to classify the metadata. I think it is
possible to enumerate some simple questions that reflect the kind of
metadata that consumers expect to have.

Then, in this case, there will be a more general requirement
"R-MetadataAvailable" [1] and from the questions more specific requirements
may be derived. Later on, we may decide if each specific requirement will
become a BP. It is also important to keep in mind that best practices
should be testable.

@Christophe, I like the idea of recommending vocabularies, but I'm not sure
if this should be part of a BP. Maybe, examples using specific vocabularies
can be used to illustrate a given BP.

kind regards,
Bernadette




2014-11-18 9:10 GMT-03:00 Christophe Guéret <christophe.gueret@dans.knaw.nl>
:

> Hi Makx,
>
> Sounds like a very  good plan! Could we maybe just recommend some
> vocabulary to use for each of those bullet point?
>
> Christophe
>
> --
> Sent with difficulties. Sorry for the brievety and typos...
> Op 8 nov. 2014 09:38 schreef "Makx Dekkers" <mail@makxdekkers.com>:
>
> All,
>>
>>
>>
>> This morning, I’ve been trying to dig up the discussion about metadata in
>> the minutes of the F2F in San José. Although it is sometimes hard to
>> understand from the log what was said and why, the impression I get is that
>> we seem to get into some fundamental discussions about what metadata is,
>> what kinds of metadata are relevant, and what metadata should be provided
>> in an ideal world.
>>
>>
>>
>> Although I do think these are very interesting topics – after all, I have
>> been participating in similar discussions for a couple of decades ;) – I
>> fear we might not reach consensus on a practical level any time soon. I
>> would suggest to take it from a different angle, and look at what kind of
>> advice publishers of data would be looking for.
>>
>>
>>
>> In my mind, the best practice for metadata should give some simple
>> guidelines on the highest level.
>>
>>
>>
>> Something like:
>>
>>
>>
>> ·         Provide as much information about the data as you possibly can.
>>
>>
>>
>> ·         Try to provide at least information about:
>>
>>
>>
>> o   What the data represents
>>
>> o   Where it is
>>
>> o   Who is responsible for it
>>
>> o   How the data is (technically) expressed
>>
>> o   What you can do with the data
>>
>>
>>
>> ·         Publish metadata with a level of quality and granularity, and
>> in a format that is expected to be useful for the intended audience. On
>> this issue, we need to acknowledge that specific applications may need
>> specific metadata that is not covered by a general standard like DCAT.
>> Maybe we can suggest that publishers use DCAT for general properties, or
>> alternatively, map to DCAT from similar properties in a domain-specific,
>> application-specific or resource-specific metadata approach.
>>
>>
>>
>> If we can work towards such a top-level set of recommendations, people
>> who feel like it can spend time – and lots of it ;) – to dig deeper into
>> defining types of metadata, identifying specific properties needed for
>> specific types of usage, building application profiles for provenance data,
>> etc., etc. This additional detail may still be in scope for the working
>> group, but maybe we can move this to a later time.
>>
>>
>>
>> And please, let’s not try to define mandatory minimum sets of metadata
>> properties. My experience is that for almost every property you declare
>> mandatory, someone can come up with a valid case where the information
>> cannot be provided. Even in the five bullets above, some publishers might
>> not have information about every category, but we should not discourage
>> publication of that data – users might not care about missing information
>> or they might have other ways to find out the bits of information that are
>> missing.
>>
>>
>>
>> Makx.
>>
>>
>>
>>
>>
>> *From:* Christophe Guéret [mailto:christophe.gueret@dans.knaw.nl]
>> *Sent:* Friday, November 07, 2014 9:23 AM
>> *To:* Data on the Web Best Practices Working Group
>> *Subject:* Re: dwbp-ISSUE-79 (metatype): Discovery vs structural
>> metadata [Best practices document(s)]
>>
>>
>>
>> Hoi,
>>
>> Digital archives also define several type of metadata (not sure how
>> many). Would it be a good idea to align this with their definitions?
>>
>> Regards,
>> Christophe
>>
>> --
>> Sent with difficulties. Sorry for the brievety and typos...
>>
>> Op 6 nov. 2014 22:22 schreef "Data on the Web Best Practices Working
>> Group Issue Tracker" <sysbot+tracker@w3.org>:
>>
>> dwbp-ISSUE-79 (metatype): Discovery vs structural metadata [Best
>> practices document(s)]
>>
>> http://www.w3.org/2013/dwbp/track/issues/79
>>
>> Raised by: Phil Archer
>> On product: Best practices document(s)
>>
>> At TPAC we made the distinction between discovery metadata and structural
>> metadata. This needs to be reflected in the BPs.
>>
>>
>>


-- 
Bernadette Farias Lóscio
Centro de Informática
Universidade Federal de Pernambuco - UFPE, Brazil
----------------------------------------------------------------------------