Hi Pascal, The reason I've asked you to forward the discussion here is because the problem seems to be a bit more general than just limited to the DBpedia German. First of all thanks to your post I noticed that our interlinking script had some bugs in it, forwarded that issue to Dimitris who discovered even more bugs, so now we should have a better Yago categorization in the country chapters.
As an answer to your specific questions > 1. why is for the German dbpedia entry[4] only taken a subset of rdf:type of > the English one[3] ? Because each internationalized DBpedia doesn't directly map yago classes to specific resource, I don't even think we could since Yago is in English(however I have no idea, maybe it would work by defining custom extraction rules and them mapping them to the corresponding English Ontology classes). We use the DBpedia Internationalization Interlinking shell script [1] which goes over the interlanguage links and and copies the Yago categorization from the matching DBpedia.org resource. > 2. why is no rdf:type reflecting that the resource[4] is (at least) also a > novel? > (I guess being typed as literature and a movie at the same time is > semantically > a contradiction, but then it would be fine to mark that contradiction in the > dataset.) Because Wikipedia contains incorrect data, as in many other cases. If you look at [2] you will see that the Wikipedia article describes the movie not the novel, however it is linked to an article in the German Wikipedia that describes the novel not the movie [3]. The previous issue results in a mismatch in the Yago categorisation in the german DBpedia endpoint, and can only be remediated by correcting the language links in the English Wikipedia. [1] http://dbpedia.hg.sourceforge.net/hgweb/dbpedia/extraction_framework/file/f143aaa9b564/scripts/shell-scripts/interlinking/interlinking.sh [2] http://en.wikipedia.org/wiki/Enchanted_April [3] http://de.wikipedia.org/wiki/Verzauberter_April On 05/18/2012 09:44 AM, Pascal Christoph wrote: > Hi *, > > (I was redirected from the German dbpedia list[5] to this mailing list). > (Btw, sorry for not having the time to further examine the paper which > explains > yago and dbpedia[1]. I am just pointing to things showing on the surface.) > > A German wikipedia resource[0] describes a novel and thus is categorizied as > literature. Now, their is a the link to "other languages", and these resources > describe the movie based on that novel. (It's wrong to link theses resources > with owl:sameAs, but of course you can only take what's there.) > The german dbpedia entry is rdf:typed using a subset of rdf:type clearly > coming from the international dbpedia[3]. Now, what makes me curious is: > > 1. why is for the German dbpedia entry[4] only taken a subset of rdf:type of > the English one[3] ? > 2. why is no rdf:type reflecting that the resource[4] is (at least) also a > novel? > (I guess being typed as literature and a movie at the same time is > semantically > a contradiction, but then it would be fine to mark that contradiction in the > dataset.) > > -o > > [0]https://de.wikipedia.org/wiki/Verzauberter_April > [1]http://www2007.org/papers/paper391.pdf > [3]http://dbpedia.org/resource/Enchanted_April > [4]http://de.dbpedia.org/resource/Verzauberter_April > [5]https://sourceforge.net/mailarchive/forum.php?thread_name=4FB3E30F.6080503%40gmail.com&forum_name=dbpedia-germany > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Dbpedia-discussion mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Dbpedia-discussion mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
