W3C home > Mailing lists > Public > public-xg-lld@w3.org > August 2011

Re: Visualizing metadata element sets in side deliverable -- quite urgent feedback request

From: Antoine Isaac <aisaac@few.vu.nl>
Date: Fri, 26 Aug 2011 15:32:37 +0200
Message-ID: <4E57A075.10005@few.vu.nl>
To: Jodi Schneider <jodi.schneider@deri.org>
CC: public-xg-lld <public-xg-lld@w3.org>
Hi Jodi,


> Do you think that [LOV] is representative, without basing it directly on our use cases? In that case, it would additionally be worth including.


Yes, even if it's much less representative (I think the size is derived from vocabularies re-using themselves, not usage in datasets) we will keep a pointer to it in the report.


> It's a shame that there aren't statistics we could easily get out of CKAN, using our LLD CKAN group (rather than the use cases) as the starting point. But I do concede that we don't have so much time!
>


In fact, it would even require data that is not fully in CKAN yet. E.g., my own http://ckan.net/package/europeana-lod has format-dc and format-ore tags, but it's not complete (there's no tag for Europeana's own vocabulary). Plus I'm quite sure that many other package curators don't know such tags exists.
In fact I'm not sure how the LOD Cloud guys could create http://www4.wiwiss.fu-berlin.de/lodcloud/state/#terms .
Perhaps by analysing the dumps, which is of course impossible for us.


> Yes, and yes. It will also need a short paragraph explaining what the tagcloud represents.
>


Of course. Thanks for the feedback!

Antoine



>
> On 26 Aug 2011, at 13:04, Antoine Isaac wrote:
>
>> Dear all,
>>
>> Among the TODOs to clean at the last moment for the side deliverable, there is one on visualizing Metadata Element Sets:
>> http://www.w3.org/2005/Incubator/lld/wiki/Vocabulary_and_Dataset#Metadata_Element_Sets
>>
>> The original idea was to show the relations between Sets and their relative importance (as indicated by their usage in individual cases).
>> The inspiration was Mondeca's Linked Open Vocabularies: we wanted to do the same kind of graph as http://labs.mondeca.com/dataset/lov/details/vocabularySpace_Library.html, but
>
> Do you think that this is representative, without basing it directly on our use cases? In that case, it would additionally be worth including.
>
>> - based on our use cases  only
>> - with links between element sets, as in http://labs.mondeca.com/dataset/lov/details/vocabulary_skos.html
>
> It's a shame that there aren't statistics we could easily get out of CKAN, using our LLD CKAN group (rather than the use cases) as the starting point. But I do concede that we don't have so much time!
>
>>
>> There was even a mock-up:
>> http://www.w3.org/2005/Incubator/lld/wiki/File:LLD-MetadataElementSetCloudMock.png
>> Unfortunately, it has obvious that the editors of the side deliverable don't have enough resources to gather the data needed to create this: Mondeca's scripts use a SPARQL endpoint, for which we'd have to create fairly rich data.
>>
>> BUT as I was about to remove the mock-up, I saw that the site Paul Walk has made in Pittsburgh was still alive: http://www.paulwalk.net/lldvis/
>> Even better, I've discovered that:
>> 1. I can easily remove from this page the value vocabularies, keeping the element sets only
>> 2. I can add new element sets, fitting the latest version of our side deliverable. In fact we can really add them in the data behind Paul's page, and attach use cases to them. But perhaps that's not needed, we can tweak the size of the tags manually...
>>
>> So I've implemented step 1 and created a tag cloud for the metadata element sets:
>> http://www.few.vu.nl/~aisaac/lld/LLDMES.html
>>
>> It's clearly not the full monty we were dreaming of, but it still gives some idea on usage of element sets.And it's really easy to make.
>> I could continue this, adding the element sets we miss. And then take a screenshot to be included in the side deliverable: I'm afraid we can't fit in the real HTML code, as it is heavily depending on scripts that are on Paul's site.
>>
>> Do you think that would be worth it? And it's not too late?
>
> Yes, and yes. It will also need a short paragraph explaining what the tagcloud represents.
>
>
> :) -Jodi
>
>
>>
>> Cheers,
>>
>> Antoine
>>
>
Received on Friday, 26 August 2011 13:30:58 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 26 August 2011 13:30:58 GMT