Reconciling data with dbpedia

Hi all, let's say I have an excel with a list of places. This is an extraction of the list: - Canada Toronto Ontario - Italy Rome Lazio - Caraibi Palm/Eagle Beach Aruba - ...

Is there a way to call dbpedia in order to associate to each place the exact dbpedia URI?

Thanks, Luca

Check out the RDF Extension for Google Refine. It has a reconciliation feature that's intended exactly for this kind of usage.

Two methods come to mind:

  • Pull all of the place names from DBpedia with a Sparql query and use the edit distance betw your terms that theirs to identify matches.
  • Use a geocoding service (like Google's) to associate each of your places with a lat/long, and then do the same with each place name in DBpedia. Then match'em up based on spatial distance.

I'd personally veer towards the second, even though it's a bit more work. Geocoding web services tend to do a good job at guessing exactly which place on earth you're referring to with a string, so you can rely on them to figure out that both "St. Louis, Missouri" and "Saint Louis, MO" refer to the same place, while "Stalouis, Montana" is somewhere else (despite what I'm guessing is a small edit distance).

I believe you can. The issue you would run into is disambiguation. Not sure what the best way is... however I believe you can use the Zemanta web service to try and find relevant articles in dbpedia. I'd start by looking at the developer API

Another way might be to search your terms on SWSE (seems to be down at the time of this writing) which has linked data references to dbpedia. Here's a link I had from a search I made note of previously.

I thank you all for your replies. Trying to figure out and choose the best way, I founf that the most Reliable solution is google :) if I try a site search on wikipedia.org domain using one of the examples reported before I can see really matching results. The first one is quite often the best one.

Google should collaborate for a reconcilitation service :)