Sunday, 1 February 2009

Repository meets the Semantic Web. Semantic Web Acquits Itself Well

I am helping to organise the Web Science 2009 conference in Athens in March, and I am putting the conference papers in a repository to generate all the paper lists for the schedules etc.

This is a new conference for an emerging discipline and it seems particularly important to be able to give an impression of the breadth of the community contribution. An obvious way to show international contribution is to plot the conference contributors on Google Earth. That's a fairy standard EPrints export demo, but I just need to know where all the contributors are (their latitude and longitude).

It's easy enough to import the affiliation, country and email for each author from the conference submission system (EasyChair) into the repository (EPrints), but there's no central service that will give me the lat/long of a university. Wikipedia has them, but there's no easy way to go reliably from the entered Affiliation data to a Wikipedia University entry. The best way that I found was to use each author's email address (or part of it!) to do a semantic web search of DBpedia for matching Universities, look up the city that the University is located in and find out the latitude and longitude of that city. It's all automated in SPARQL, so it's pretty efficient (now that I've learned about DBpedia and SPARQL that is!) It may have been just as quick to do it by hand from Wikipedia, but where's the fun in that?

1 comment:

  1. I tried the Tinderbox geocoder, which uses Google Maps

    It does surprisingly well on some test cases: "Harvard University", "Miami University of Ohio", "University of Southampton"