hi all, actually DBnary is an up to date approach. Maybe Gilles can post something about the current state.
we didn't get funding for the wiktionary extraction and Jonas didn't want to become my phd student :(
we really hope to create a huge mesh of all open dictionary date, soon.
Sebastian
On 27 November 2014 3:44:34 PM GMT+01:00, Karl Bartel karl42@gmail.com wrote:
Hi Dimtiris,
it's excellent to hear that someone is still taking care of this project! It seems to be the most promising extraction approach to Wiktionary and I was afraid it might have been abandoned completely.
Can you please try this link for editing? http://wiki.dbpedia.org/Wiktionary/edit
Thanks, that works. I will update it later this week in those places where I'm confident that my basic but current knowledge is better than the existing outdated text.
Can you try to run the project from github and tell me if you run
into
errors again? https://github.com/dbpedia/extraction-framework
After figuring out how to get everything to run, it mostly worked well. I have extracted triples from the current English and German dumps and loaded them into a local fuseki server for some basic testing. A lot of data is there, but there seem to be at least some extraction problems. The most obvious thing is that the hasTranslation triples are not generated. I didn't find the time to understand the translation helper, yet. So I'm not sure whether that is just a minor syntax change in the Wiktionary or something is really broken. Is there anything I should know about the translation extraction? Why is a helper needed for it, instead of an ordinary template tag with result triples?
Karl
Wiktionary-l mailing list Wiktionary-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wiktionary-l