Project rdf:bout
Steve Bennett
> Note that in a few of the XML files you've attached in issues, I noticed
> that you've stripped the rdf:about, leading to all projects submitting
> containing:
>
> <Project rdf:about="http://registry.oss-watch.ac.uk/">
>
> In the doapcreator javascript the rdf:about is generated (line 349) as:
> http://registry.oss-watch.ac.uk/projects/shortname
> where shortname is the 'Unique project tag' in the UI.
Just wondering about this - should the rdf:about be a resolvable URL,
or just any URI? As a rule, we don't have any web address that
represents each project - that's what Simal is for! But I can't
actually make the URI that of the project address in Simal, because I
don't have the project ID at the time of import. Suggestions?
Steve
Ross Gardler
> On 11 March 2011 12:43,<si...@googlecode.com> wrote:
>> Note that in a few of the XML files you've attached in issues, I noticed
>> that you've stripped the rdf:about, leading to all projects submitting
>> containing:
>>
>> <Project rdf:about="http://registry.oss-watch.ac.uk/">
>>
>> In the doapcreator javascript the rdf:about is generated (line 349) as:
>> http://registry.oss-watch.ac.uk/projects/shortname
>> where shortname is the 'Unique project tag' in the UI.
>
> Just wondering about this - should the rdf:about be a resolvable URL,
> or just any URI?
I always thought it just needs to be a URI, but Hugh recently said:
"In short, you should have URIs for each project that somehow return RDF
(either with a 303 or a hash URI)"
Hugh was not talking about RDF requirements but "best practice" in
linked data. You can see his full comment at
http://code.google.com/p/simal/issues/detail?id=402
> As a rule, we don't have any web address that
> represents each project - that's what Simal is for! But I can't
> actually make the URI that of the project address in Simal, because I
> don't have the project ID at the time of import. Suggestions?
If you provide data without an rdf:about Simal will create a URI for
you. In theory when you import data about that project again, in the
future, it will detect they are the same project and use the same URI
again. However, it currently uses the project homepage to detect
duplicates so this won't help in your case.
I recommend that you either create a URI (that satisfies Linked Data
best practice) or you help us improve the project duplicate detection.
Ross
Sander van der Waal
> On 11 March 2011 12:43, <si...@googlecode.com> wrote:
>
>> Note that in a few of the XML files you've attached in issues, I noticed
>> that you've stripped the rdf:about, leading to all projects submitting
>> containing:
>>
>> <Project rdf:about="http://registry.oss-watch.ac.uk/">
>>
>> In the doapcreator javascript the rdf:about is generated (line 349) as:
>> http://registry.oss-watch.ac.uk/projects/shortname
>> where shortname is the 'Unique project tag' in the UI.
>>
>
> Just wondering about this - should the rdf:about be a resolvable URL,
> or just any URI?
It should be a resolvable URI but the way I set this up was for the URI
to be resolvable to the running Simal instance.
> As a rule, we don't have any web address that
> represents each project - that's what Simal is for!
That means there's a running instance of Simal on a certain host name.
Can't you use that host name for your project's rdf:about?
> But I can't
> actually make the URI that of the project address in Simal, because I
> don't have the project ID at the time of import. Suggestions?
>
I guess the way to resolve that would be to generate a temporary ID for
your project. Note that there is no consistent way yet for managing the
project ID in the interface. So this needs more work in any case.
Sander
> Steve
>
>
Steve Bennett
<sander.v...@oucs.ox.ac.uk> wrote:
> That means there's a running instance of Simal on a certain host name.
> Can't you use that host name for your project's rdf:about?
Yeah, that gives the host, but what about the rest? Say our domain is
simal.ands.org.au. Ideally we would have rdf:about's like
simal.ands.org.au/project/PROJCODE. Perhaps I should look at adding
something like that to the rest interface - look up by project code.
But I guess even if it doesn't (yet) resolve it's not a huge problem.
To answer Ross's remark:
>I recommend that you either create a URI (that satisfies Linked Data best practice) or you help us improve the project duplicate detection.
I don't think satisfying linked data best practice is feasible for us
in the short term - I need to be focused on getting something up and
running. Being a good semantic web citizen is a much lower priority
(unfortunately). The second option is a lot easier. It depends a bit
how we resolve the related issue of doap:shortname not existing. Once
we have a solution to that, then we would make that another (higher
priority) duplication detection mechanism: if two projects have the
same code, they are the same project.
Now the question on that front is whether we invent some new
ANDS-specific field (ands:projectcode), use a generic identifier
(dc:identifier) or invent a Simal-specific field (simal:projectcode).
In the latter two cases we would need some sort of namespacing, I
think.
Steve
Ross Gardler
On 15 Mar 2011, at 01:22, Steve Bennett <ste...@gmail.com> wrote:
>
>> I recommend that you either create a URI (that satisfies Linked Data best practice) or you help us improve the project duplicate detection.
>
> I don't think satisfying linked data best practice is feasible for us
> in the short term - I need to be focused on getting something up and
> running. Being a good semantic web citizen is a much lower priority
> (unfortunately).
Hmmm... That assumes it would be hard to do. But if you want to make that assumption then fair enough.
> The second option is a lot easier. It depends a bit
> how we resolve the related issue of doap:shortname not existing.
I'm not sure why shortname has anything to do with this. A project can have more than one shortname, so is not useful here.
simal:id is what is used to identify a project.
> Once
> we have a solution to that, then we would make that another (higher
> priority) duplication detection mechanism: if two projects have the
> same code, they are the same project.
What is "code"? If you mean simal:id then that is already used for duplicate detection.
> Now the question on that front is whether we invent some new
> ANDS-specific field (ands:projectcode), use a generic identifier
> (dc:identifier) or invent a Simal-specific field (simal:projectcode).
> In the latter two cases we would need some sort of namespacing, I
> think.
Why another property at all?
simal:id is what is currently used in the URI, unless you provide a URI on submission. That property only needs to be unique on each instance of Simal, when it's imported a world unique identifier is created.
Ross
>
> Steve
>
> --
> You received this message because you are subscribed to the Google Groups "Simal contributors" group.
> To post to this group, send an email to simal-con...@googlegroups.com.
> To unsubscribe from this group, send email to simal-contribut...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/simal-contributors?hl=en-GB.
>