[ckan-discuss] Editing package relationships (was: Re: [ckan-dev] Package Resources Proposal)

David Raznick kindly at gmail.com
Thu Feb 3 21:44:10 GMT 2011


On Wed, Feb 2, 2011 at 11:29 PM, William Waites <ww at styx.org> wrote:

> * [2011-02-02 23:17:41 +0000] David Raznick <kindly at gmail.com> écrit:
>
> ] Currently the only active relationships are of the linked-to type.   I
> think
> ] they are meant as general expressions that the data within the rdfs they
> ] contain are linked. I can not prove this though..
> ]
> ] Are there any linked data spiders to do these links automatically?
>
> as i understand it (cygri or anja may correct me) they initially
> attempted to use the package relationships and then found that it
> was too cumbersome and made a convention for using the extras.
>
> Very interesting.

It appears the convention is not always followed. i.e
http://ckan.net/package/bibbase (there are some that do not start with
"links").  At one point I may run through all the keys in the extras table
to see if they are a package names.   I may also redo the graph knowing that
extras are the more definitive source of links.

Should we adopt this convention? So as to make  links:package clickable in
the same way shortcuts in trac work?.
This kind of 'content' linking seems the right choice here.

these numbers are estimates of the number of links from one dataset
> to another, as the size of the dataset is also an estimate (there
> is a tangential debate here about what sizes and counts of these
> types mean in the presence of rules).
>
> because the datasets in question can be very large, hundreds of
> millions of triples and because some of them are not downloadable
> in bulk and yet others are countably infinite it isn't feasible
> at this time to actually calculate these numbers. so afaik no
> spiders to analyse the data at this level.
>


It still seems a shame for smaller datasets that this could not be done.
Not being able to do this seems to defeat the object of linked data in my
eyes.
I do not think accurate counts are necessary.  In the same way a search
engine does not need to know a count of all the results.   In a perfect
world I would like also like to see a sample of the triples (or even better
a local graphs) of what caused the link.


> cheers,
> -w
> --
> William Waites                <mailto:ww at styx.org>
> http://eris.okfn.org/ww/         <sip:ww at styx.org>
> F4B3 39BF E775 CF42 0BAB  3DF0 BE40 A6DF B06F FD45
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.okfn.org/pipermail/ckan-discuss/attachments/20110203/36784c31/attachment.htm>


More information about the ckan-discuss mailing list