[euopendata] A data catalog in dcat mapped to Eurovoc...

[Peter Krantz]

2011/2/6 Friedrich Lindenberg <friedrich.lindenberg at okfn.org>:
>
> Just to verify: does this support the more eccentric features of
> AtomPub, such as paging? Looking at it now, it seems to me the number
> of records is smaller than what I think is in your catalog.

It should contain all catalog items (around 100). I will implement
feed paging when the number of items increase significantly.

>
> 1) It seems to me that "xmlns:catalog="http://data.gov/catalog/">" was
> left in by accident.

Damn. Coding by cut'n'paste. Added to fixlist.

> 2) You're specifying formats (in a nice way, btw!) on a per-dataset
> level. My previous understanding had been that this would be a
> property of the distribution, e.g. you'd have a Distribution accessURL

Ah OK. I see I missed that it has domain Distribution. I'll set up a
distribution for each format.


> www.foo.org/test.xml with format "XML" and a second one to
> www.foo.org/test.json with format "JSON". I think we should find a
> shared convention on this - it just makes re-use so much easier.

It seems that the definition indicates that it isn't necessarily a
direct download link. It could be a link to a splash page where the
data may be found.


>
> This is extremely useful. How do you align your datasets to Eurovoc -
> using some mapping of keywords or manually?

I did manual mapping for keywords that already were used for tagging
catalog entries. Eurovoc has improved significantly and is now
available in SKOS with proper URI:s it seems. I looked them up on the
Eurovoc website but it looks like it is possible to download the
entire thing in bulk.

Another question: the catalog contains datasets produced by agencies
that only are available under substantial commercial terms. Is there a
way to mark them in dcat to make it easier to exclude them from
indexing? I raised this issue a while ago, but I remember getting
feedback that CKAN will encompass that type of data too.

Regards,

Peter