- From: Hogan, Aidan <aidan.hogan@deri.org>
- Date: Wed, 13 Apr 2011 23:21:43 +0100
- To: "Bernard Vatant" <bernard.vatant@mondeca.com>
- Cc: "Linking Open Data" <public-lod@w3.org>
> So tonight I would turn my question otherwise : Among those millions of > FOAF profiles, how do I discover those of which primary source is their > primary topic, expressing herself natively in FOAF, vs the ocean of > second-hand remashed / remixed information, captured with or without clear > approbation of their subjects, and eventually released in FOAF syntax in > the Cloud ... Can't think of a single definitive solution that would work for all scenarios, but there are four or five heuristics you could think of combining: - filter all big FOAF exporters (there are not *so* many and you already have a good list to start from); - look for FOAF files with the FOAF-a-Matic generatorAgent - look for FOAF documents which have a dc:creator/foaf:maker relation to a person in the document (or a foaf:made in the other direction); - look at the "entropy" of URIs which are the objects of FOAF knows relations... e.g., are they commonly in different "namespaces"? - look for rare "geek" properties like foaf:tipjar, foaf:myersBriggs, or foaf:dnaCheckSum - ... ...there are many other tell-tale signs you could look at. I guess it depends on what kind of precision/recall you need, but these should get you a good bit of the way. Good hunting! Aidan
Received on Wednesday, 13 April 2011 22:22:16 UTC