Re: Europeana API as an evidence for the DWBP

Hi Annette,

Thanks for your message and suggestions! Please, find my comments below.

I fear that our method of collecting implementation information is making
> submissions burdensome and leading us to the wrong kind of evaluation.


I don't agree :( We had a lot of discussions about this and we tried to
explain how to collect the evidences using both forms. Maybe, the
explanation was not clear. See more comments below.

These discussions of whether a particular site is a good one are not what
> we're after. What we really want to collect is not evaluations of whole
> sites but a very short list of definitely compliant sites for each BP.


I agree with you! This is also our idea behind collecting
evidences/implementations for the DWBP. Our main goal is to to show that it
is possible to implement each one of the proposed BPs.  To do this we can:

(i) ask people to implement new datasets using one or more of our BPs to
show that it is possible or not to implement it; and
(ii) show that existing datasets implemented our BPs, i.e, it shows that
our proposal is possible to implement;

After the call with the director, we also agreed to collect evidences from
existing datasets to show that our BPs are in fact 'best practices', i.e.,
they are already considered by known organizations and datasets/portals.

IMO, we shouldn't see this is an evaluation of a specific dataset or
portal. It is important to remember that we are evaluating the DWBP and not
the datasets.



> So we don't really need to ask people to fill in a complete evaluation of
> each site they offer as an example, and I doubt that we want to use the
> fact that something is "partially compliant" for a single BP at all.


As we explained before, an evidence doesn't need to cover the whole set of
BPs. The form shows the complete list of BPs, but it is up to one who
provides the evidence to choose the BPs that are suitable or not. The
process doesn't need to consider an evaluation of each BP for each evidence.

The idea of being partially compliant is more suitable when someone is
trying to implement a new dataset using the DWBP (case (i) above). For the
other case (case (ii)), it is more important to show which BPs were
considered.


> I would suggest a form that simply asks for the site info and then lets me
> associate that with a BP from a drop-down menu. It would be handy for the
> form to return, in addition to verification of the submission, another
> input form with the site info prefilled, so that I could choose another BP
> and submit that right away, or just call it a day. That would relieve the
> pressure to try and fill out a painfully detailed evaluation.
>

I'm afraid that now it's too late to change the way of gathering evidences
:( I think the way that we made the form available in [1] is sufficient to
collect all the information needed in both cases, i.e., creation of a new
dataset or using an existing dataset as an evidence.

After our last meeting, we also agreed to include other types of evidences,
like guidelines and blogs. In this case, the form can be adapted. Please,
feel free to make the changes that you think are necessary to collect this
new type of evidence.

kind regards,
Bernadette

[1]
https://docs.google.com/spreadsheets/d/1JE5pDy9YCu9eafQv50JJ3SauK4Jq1QmagV-GCDNY24E/edit?usp=sharing


>
> -Annette
>
>
>
> On 10/30/16 3:39 PM, Antoine Isaac wrote:
>
>> Hi Bernadette,
>>
>> Thanks. In fact your effort shows good and bad points:
>> - it's hard to collect all the details when one is not very close to the
>> dataset
>> - it's good that you contact people and nag them ;-)
>>
>> I've tried to add some comments to the form. And here's where the lack of
>> time and generality of BPs hits. For many I was not sure that what I was
>> thinking of would change a 'fail' into a 'pass'. So I've just put comments
>> ina  new column, letting you judge - though for some BPs I'm quite
>> affirmative that the API would pass.
>>
>> Also, I've generalize my response to include the linked data service. It
>> becomes difficult to separate the LOD from a more 'traditional' API when
>> one is built around the other, and both are on the same namespace. And even
>> if our LOD service is less mature, we still intend it to be a recognition
>> that some of the LOD recommendations are indeed BPs that we want to follow,
>> and thus can be counted as 'implementation' (in the wider sense) of the
>> specified BPs.
>>
>> So maybe it is better then to consider a wider 'Europeana Data service'
>> item than just the 'traditional' API.
>> Actually the various services we have at Europeana can also be seen as a
>> token that some of us at Europeana do agree with some (not all!) of the
>> points raised in the blog posts Pieter just sent. Even if that's another
>> story - the point right now is that it's much better to consider our
>> complete data offer not just one API.
>>
>> Best,
>>
>> Antoine
>>
>> On 27/10/16 20:19, Bernadette Farias Lóscio wrote:
>>
>>> Hi Antoine,
>>>
>>> I hope everything is fine with you! We are still collecting evidences
>>> for the DWBP and I was considering to include the Europeana API as an
>>> evidence.
>>>
>>> I was taking a look on the Europeana Labs site and I made a first report
>>> about the API [1]. It would be great if you could take a look! Please, let
>>> me know if you agree with the evaluation and feel free to complement or to
>>> make changes.
>>>
>>> Feel free also to include other evidences.
>>>
>>> Thanks a lot!
>>> Bernadette
>>>
>>> [1] https://docs.google.com/spreadsheets/d/1mRVp8Vdudepk68AjQbhN
>>> oZnLU-0-vH_4fVug_J-hCxo/edit?usp=sharing
>>> [2] http://labs.europeana.eu/
>>>
>>> --
>>> Bernadette Farias Lóscio
>>> Centro de Informática
>>> Universidade Federal de Pernambuco - UFPE, Brazil
>>> ----------------------------------------------------------------------------
>>>
>>>
>>
>>
> --
> Annette Greiner
> NERSC Data and Analytics Services
> Lawrence Berkeley National Laboratory
>
>
>


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