« Python 3.1.1 | Main | Bibliographic Statistics 2009 »



This is really nice Thom!

One comment on the FOAF, though: the foaf:name property is really for the display form of the name, but you're using it for effectively a semi-structured representation of the sort form. Is this something you might work on over time?


I see the VIAF FOAF as primarily a place holder for what might be done in the future, so we're open to suggestions on how it could be improved, especially from someone that would make use of it.


Patrick Danowski

Hi Thom,
Can you provide the ID Numbers of the Person in the different Authority files in the RDF? Or a possibility to get to the page if i know for example the PND Number of the national library. The VIAF is not very much used this would it make easier creating links to VIAF from other sources like the German set of dbpedia (i a lot of wikipeda articled the PND Number is inside.

Just a few ideas for improvement.

Response: That sounds like a reasonable thing to do. Any idea how you would like the RDF to look?

Joachim Neubert

Hi Thom, hi Ralph,

It's great to see VIAF taking the way to linked data. Millions and millions of well-defined entities for author disambiguation are a real treasure, for bibliographic applications as well as for the Linked Data Web as a whole.

I played with your website using the URIs you gave and had a conversation with Richard Cyganiak from http://pedantic-web.org about it. Richards hints had proved very helpful when I built my own Linked-Data application (at http://zbw.eu/stw). He pointed out an issue I observed on viaf.org but could not explain. When asking

curl -ILH "Accept: application/rdf+xml" http://viaf.org/viaf/36978042.rwo

I got a 303 redirect and then a 404. The reason for this seems to be that your web server doesn't support HTTP HEAD requests correctly (GET requests, with curl -iLH, work). This should be fixed, since clients with an own cache often use HTTP HEAD to see if a resource has changed.

Another suggestion about naming the resources: It's a bit a miracle what ".rwo" may denote. In my ears it sounds quite application dependent, and it could be much clearer to use an URI like


to designate the person, and a 303 redirect to the actual representation.

Richard raised the question if content negotiation is helpful at all in your situation, because clients normally will not address unimarc or marc21 in HTTP accept headers, so client programmers would probably use self-built URIs anyway. In his page "Frequently observed Problems on the Web of Data" (http://pedantic-web.org/fops.html) he says a lot more about that topic.

One last point: As of now, the XML representation is much richer than the RDF representation. The latter basicly provides links to other representations, but is currently in no way equivalent to them, which could be missleading to clients. But I take your words for that you plan to enrich the RDF representation with all the valuable data you already have in XML - which would be really, really great.

All the best for your work.

Cheers, Joachim

The comments to this entry are closed.

My Photo

May 2015

Sun Mon Tue Wed Thu Fri Sat
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30