Cool URIs in VIAF

Since VIAF started creating merged clusters we have tried to be careful about our URIs.  Since the system is evolving, this has been a challenge, but I think we've done fairly well. Our first URIs were very simple:

http://viaf.org/95216565 (for Chekhov)

That URI was created in 2009 and still works, although it now redirects you to:

http://viaf.org/viaf/95216565

The 'extra' viaf in the URI gives us more flexibility in constructing new services and resource sets within the viaf.org domain.

That's not to say there haven't been changes to the cluster. A large group of Chekhov source records were added to the cluster in late 2009, more in 2010 and a couple of source records have been deleted. Since the contents change as clusters merge and split, we have to decide which source records are associated with which VIAF cluster. Originally we optimized changes to minimize the number of clusters which disappear or are created. This was important when we did not have a mechanism in place to redirect from one VIAF ID to another, but had a side effect of moving more source records from one cluster to another than we liked.

We now have a mechanism to redirect outdated URIs to the current one. Using another Russian example, the main Stalin cluster at one time resided at:

http://viaf.org/viaf/95236870 

that URI redirects to the current Stalin cluster:

http://viaf.org/viaf/101787139

We now optimize for minimum movement of source records between clusters, which gives us more stable clusters.

Although most discussions of cool URIs do not directly address it, another aspect of coolness is how stable the information is on the page the URI eventually resolves to. We have made a number of changes, especially in the RDF representation of the clusters. The native XML cluster records have been more stable, but even there we occasionally add or change fields, and as we expand to new types of names we will have to make more changes. This is all new territory for us, so it is hard to promise there won't be more changes, although we have no plans for substantial changes to the current RDF, and we will continue to use all the RDF tricks available to make existing URIs within the RDF stable.

Sometime soon we will do another blog post about URI patterns in VIAF, but here is one than anyone interested in the internals of VIAF should be aware of:

http://viaf.org/viaf/search

That gives a formatted display of the SRU Explain response showing all the indexes and allowing searching and browsing at quite a low level.  In fact, you too could build an interface to VIAF which replicates (or improves!) the current Web browser interface you see at viaf.org.  Except for a few URL rewrite rules embedded in our Tomcat server, everything is exposed, including the XSLT style sheets that transform the XML to HTML for display.

--Th

Comments

[Richards1000]

Many thanks for this. Would you please tell us the SPARQL endpoint for VIAF?

[Thom]

Sorry, we don't have a SPARQL endpoint for VIAF.

--Th

[Thomas Berger]

The Stalin redirection example and the search page currently appear to be broken.

(Update: this is now fixed. Thanks, --Th)