The score that I'm using is derived from the Wikipedia pagecounts data http://dumps.wikimedia.org/other/pagecounts-raw/
which is (sort of) a direct sample of people's interest in the topics, so it's very different from the PageRank data computed by Andreas Thalhammer for wiki-en, see http://people.aifb.kit.edu/ath/ Anyway, DBpedia extracts the redirects and even does the transitive closure on them, so it's not a difficult problem (in principle) to resolve the redirects using DBpedia data and then join against Wikidata. On Mon, Feb 17, 2014 at 12:12 PM, Klein,Max <[email protected]> wrote: > Hi Paul, > > I don't know exactly how your "importance" ranking is defined, but it seems > like you are making some "PageRank-y" inferences based on the incoming links, > and incoming redirects. > > Putting aside the unprovedness of redirects conferring PageRank for a moment. > > Your question as I understand it, is how to find all the redirects, of all > the sitelinks, of a given WikidataQID. > In general I don't think this is easy, but a complicated solution may not be > too difficult to compute. Redirect pages are just pages in the main namespace > that have the "#REDIRECT" syntax. That means you can find them by > transclusion, and some filtering. Like in this API call > > http://en.wikipedia.org/w/index.php?title=Special:WhatLinksHere/Deadpool&hidelinks=1&hidetrans=1 > > (It also seems that this tool will do it: > http://toolserver.org/~dispenser/cgi-bin/rdcheck.py?page=Deadpool ) > > But if you were going to look at ALL the redirects anyway, another way to do > this might be to start looking at redirects and matching them into your > preexisting targets. You could do this using the database replicas on > Wikimedia labs and the redirect table > https://www.mediawiki.org/wiki/Manual:Redirect_table > > Maximilian Klein > Wikipedian in Residence, OCLC > +17074787023 > > ________________________________________ > From: [email protected] > <[email protected]> on behalf of Paul Houle > <[email protected]> > Sent: Friday, February 14, 2014 10:21 AM > To: [email protected] > Subject: [Wikidata-l] Reconciliation of Wikipedia usage data with Wikidata? > > Lately I've merged Wikipedia usage information with DBpedia > information to produce a subjective importance score: > > https://github.com/paulhoule/telepath/wiki/SubjectiveEye3D > > That particular product has two layers of processing that can be done > independently: one is averaging over time, the other is reconciling > the {project,page URI} tuples to concepts. In the case of DBpedia, > the main thing that happens is that importance is summed up for all of > the redirects that point to a particular concept. > > The product linked above was built with the English Dbpedia, but it > ought to be easy to do the same thing with a different ?lang Dbpedia, > and get products that reflect the point of view of different ?lang > zones. > > I don't see redirect data in Wikidata, so it seems that redirects > should be processed against DBpedia of all languages and then these > should be merged with the contents of the "Wikipedia pages linked to > this Item" that I see on a page like this: > > http://www.wikidata.org/wiki/Q1631090 > > does this make sense? What's the easy way to get bulk access to that > information? Who else wants to see this happen? > > > > -- > Paul Houle > Expert on Freebase, DBpedia, Hadoop and RDF > (607) 539 6254 paul.houle on Skype [email protected] > > _______________________________________________ > Wikidata-l mailing list > [email protected] > https://lists.wikimedia.org/mailman/listinfo/wikidata-l > > > _______________________________________________ > Wikidata-l mailing list > [email protected] > https://lists.wikimedia.org/mailman/listinfo/wikidata-l -- Paul Houle Expert on Freebase, DBpedia, Hadoop and RDF (607) 539 6254 paul.houle on Skype [email protected] _______________________________________________ Wikidata-l mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/wikidata-l
