W3C home > Mailing lists > Public > public-dwbp-wg@w3.org > March 2015

Re: document biased toward linked data practices

From: Augusto Herrmann <augusto.herrmann@gmail.com>
Date: Wed, 18 Mar 2015 22:56:23 -0300
Message-ID: <CAH9TSzO9ySM_wOwFVzAY4FsbTT_7b=cWGc=WJPqypE7Vjk+7Jg@mail.gmail.com>
To: Public DWBP WG <public-dwbp-wg@w3.org>
+1 to Carlos, Eric and Annette.

In my experience with open data projects, the feasibility of doing Linked
Data is rare, that is in the few cases this is even a consideration.

The fact is that modelling your business domain into a vocabulary has a
cost. Even if it has a good ROI in the long run for the benefits in data
integration and interoperbility, many resource starved OD initiative simply
lack the resources for going that extra mile.

I agree with Carlos that RESTful APIs are great examples of data deeply
ingrained on the web, especially if they make good use of hypermedia
(HATEOAS). The fact that search engines do voraciously index each
individual resource in any such APIs is a testament to that.

Actually I'm quite surprised to see this issue being discussed this far
into the WG lifetime. I remember taking part in the initial conference
calls and non-linked data seemed to me to have always been in scope -
besides what's already written on the WG charter as Carlos and Annette
mentioned.

My 2c.

Augusto

On Wed, Mar 18, 2015 at 9:22 PM, Annette Greiner <amgreiner@lbl.gov> wrote:

> +1 to Carlos. I would not like to see this group become focused only on
> LOD. I think the charter[1] is clear that we are to consider more than
> that.The preamble states,
>
> "There are disparities between different developers too: for many, data
> means CSV files and APIs, for others it means linked data and the two sides
> are often disparaging of each other.”
>
> The mission is also clear, especially point 3:
>
>    1. to develop the *open data ecosystem*, facilitating better
>    communication between developers and publishers;
>    2. to provide *guidance to publishers* that will improve consistency
>    in the way data is managed, thus promoting the re-use of data;
>    3. to *foster trust in the data* among developers, whatever technology
>    they choose to use, increasing the potential for genuine innovation.
>
> -Annette
>
> [1] http://www.w3.org/2013/05/odbp-charter
> --
> Annette Greiner
> NERSC Data and Analytics Services
> Lawrence Berkeley National Laboratory
> 510-495-2935
>
> On Mar 18, 2015, at 4:24 PM, Carlos Iglesias <contact@carlosiglesias.es>
> wrote:
>
> Hi,
>
> On 15 March 2015 at 22:58, Makx Dekkers <mail@makxdekkers.com> wrote:
>
>> All,
>>
>>
>>
>> I wasn’t able to be on the call so I am not entirely sure in what context
>> Yaso made this comment, but I have been thinking along the same lines. It
>> seems to me that the current best practices try to take a fairly general
>> view, and maybe that is not good.
>>
>>
>>
>> If we try to define best practice for any type of data and any type of
>> technology, we’ll end up in very general statements like “provide metadata”
>> and “provide data in open formats”. How useful is that? How many people in
>> the world are going to say: o gosh, I hadn’t thought of that? I’d say
>> no-one.
>>
>
> In my experience I'd say many, otherwise we should be currently seeing
> more metadata and open formats in practice and that's not happening.
> There is nothing wrong with BPs being quite simple and evident. The good
> thing of BPs is precisely their guiding and reference character.
> In addition, there is nothing preventing us from going deeper through
> implementation techniques with specific technologies.
>
>
>> For example we now say in Best Practice 7: Provide data provenance
>> information: Use the Provenance Ontology [PROV-O] to describe data
>> provenance. Great, but what people really want to know is, how? And they
>> want to see how others are using PROV-O in practice. Or in Best Practice 3:
>> Use standard terms to define metadata: Metadata is best provided using RDF
>> vocabularies. There is nothing actionable in that advice, which means that
>> no-one is going to do anything with it, unless they already know how to do
>> that.
>>
>
> Then you can provide specific implementation techniques for those.
>
>
>> Maybe it would be more useful if we did indeed focus on Linked Open Data
>> – in some of the work that I have done, I noted that best practices for LOD
>> is something that people are screaming for.
>>
>
> Yes, asking for LOD support is usual when you work on a LOD project or
> environment and quite the opposite when not. Have been working on around 50
> different OD projects during the last 6 years and I think that maybe just
> 10% asked for LOD techs. As another example I have recently participated in
> the evaluation of 300+ proposals for OD based businesses and only roughly
> between 2-3% were planning to be using LOD at any extent. In general I
> noted that LOD is something people is not usually screaming for except in
> some specific scenarios.
>
> Maybe we should limit this work to cover advice for publishing tabular
>> data using the DataCube vocabulary and how to use DCAT for that kind of
>> datasets, with good examples of existing applications and Application
>> Profiles of DataCube and DCAT, with additional advice on when and how to
>> use PROV, VOID, VOAF – again with good examples from existing
>> implementations to show how it can be done.
>>
>
> This looks more like a quite specific guide than a best practices
> document, no?
> I think this would be a great idea as an additional WG note with an
> implementation example, but not as a replacement of the BPs themselves.
>
>
>>  So in summary, I think that the more specific these best practices are,
>> the more useful they are going to be. I understand this is completely the
>> opposite of what Carlos was arguing, but I don’t think people are going to
>> be excited about general advice.
>>
>
> I think some of the most popular recs at W3C in the past were such
> "general advice" e.g. WCAG, MWBPs...
> I think that making BPs generic and tech neutral is a BP itself.
> I think that having good generic BPs is also compatible with much more
> specific advice (in the form of implementation techniques).
> I think we should be only focusing on LOD if we agree first on modifying
> the charter and the document name and scope.
>
> Best,
>  CI.
>
>
>>
>> Makx.
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> *De:* yaso@nic.br [mailto:yaso@nic.br]
>> *Enviado el:* 13 March 2015 15:30
>> *Para:* Public DWBP WG
>> *Asunto:* document biased toward linked data practices
>>
>>
>>
>> Hi folks,
>>
>>
>> About what I said today at the end of the call:
>>
>> If we can't think in use cases where Data on the Web is not also Linked
>> Data, shouldn't we agree that this Best Practices Document can and need to
>> be biased towards Linked Data Best Practices Document?
>>
>> The BPs doc says at the intro: "The best practices described below have
>> been developed to encourage and enable the continued expansion of the Web
>> as a medium for the exchange of data."
>>
>> Imho, it closes the issue raised [1], helps us to decide about open
>> issues [2] and make things more clear about the scope of the deliverables -
>> and reinforces what phil said today about the "and if you don't want to use
>> it then don't complain" :-)
>>
>> Particularly, I think that we should keep our mind open, even that this
>> is to think in situations whether there can be data on the web that is not
>> linked data (not trivial, if not impossible?). Somehow this is connected
>> with conversations that we left behind, as well as the conversation about
>> protocols, for example. Maybe a note of the working group...
>>
>>
>> Salut,
>> Yaso
>>
>> [1] http://www.w3.org/2013/dwbp/track/issues/144
>> [2] http://www.w3.org/2013/dwbp/track/issues/open
>>
>
>
>
> --
> ---
>
> Carlos Iglesias.
> Internet & Web Consultant.
> +34 687 917 759
> contact@carlosiglesias.es
> @carlosiglesias
> http://es.linkedin.com/in/carlosiglesiasmoro/en
>
>
>
Received on Thursday, 19 March 2015 01:56:52 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 19 March 2015 01:56:52 UTC