Hi, Thank you for your reply. I don't think you need to host the big indexes, as long as the documentation is clear on how to build them.
Can you give me some input on how the configuration for the 50G index would be like? I've been trying to build one but the max I got was 7G and for some reason it only worked once following the same exact steps. I am willing to try to create a docker container to build the big index and obviously make it available to everyone, but I need to be able to do it myself first and in a way that I can reproduce it. As for the version of dbpedia, I think that can be solved by being able to work with DBPedia current. If it works once, it should always work with the most recent version available. Regards, Antero On Fri, 25 Mar 2016 12:05 pm Rupert Westenthaler, < rupert.westentha...@gmail.com> wrote: > Hi Pooja, > > I agree with Antero. > > Regarding DBPedia indexing: We had DBPedia indexes (up to version 3.8) > but the server used to download those had a hardware breakdown and was > not replaced afterwards. I think I still have full indexes of 3.8, but > I never built one for a newer version of DBPedia > > ATM I am not eager to build a new DBPedia index as the current dataset > is ~1 year old (April 2015) and I expect a new version to be released > soon. But even If I would build a new index I would still not have a > server to host the files that could be (depending on the configuration > form 5 to 50GByte in size). > > best > Rupert > > > On Tue, Mar 22, 2016 at 11:37 AM, Antero Duarte <a.fduar...@gmail.com> > wrote: > > Hi there, > > > > When you first start stanbol you get a small dbpedia index, that's why > you > > don't get the 2nd or 3rd tier cities/towns. When you use dbpedia > spotlight, > > you connect to an external service that probably uses a bigger index than > > the 43k entities, that's why you get more data but not everything. You > can > > follow the documentation to create a bigger index, start with these > links: > > > > > https://stanbol.apache.org/docs/trunk/customvocabulary.html#building-full-local-indexes-with-the-entityhub-indexing-tool > > > > http://svn.apache.org/repos/as > > > f/stanbol/trunk/entityhub/indexing/genericrdf/src/main/resources/indexing/config/indexing.properties > > > > Also: > > > https://stanbol.apache.org/docs/0.9.0-incubating/customvocabulary.html#examples > > > > But a lot of people seem to be struggling with this, can I suggest that > > someone that doesn't struggle with this and is successful to build bigger > > indexes goes through the documentation and updates it to both index > dbpedia > > current (should be applicable to future versions or at least 3.9, maybe > use > > this script <https://github.com/apache/stanbol/pull/3>) and remove older > > parts of the documentation that no longer apply to the current version of > > Stanbol? A lot of the users are struggling with this, myself included, I > > was able to build a bigger index, but it doesn't work every time and I > > can't understand why. > > > > Best Regards, > > Antero Duarte > > > > On Tue, 22 Mar 2016 at 10:17 Pooja H Bavishi < > > pooja.bavi...@iet.ahduni.edu.in> wrote: > > > >> Hello, > >> > >> I noticed that apache stanbol is able to recognize famous indian cities > >> like Bangalore, Delhi etc . but when I give a tier 2 or tier3 city or > town > >> name it is not recognizing. though the depedia has entries for such > tier 3 > >> cities (such as Bijapur, Gulbarga etc).I tried with Dbpedia spotlight > >> engines (dbpspotlightcandidates and dbpspotlightspot) though the Enitity > >> recognition has improved but I am still not getting the locations(lat > and > >> long) which is present in the dbpedia link generated by the stanbol ( > >> http://dbpedia.org/page/Bijapur) is there any way I could get the geo > >> mapping from stanbol for any location that is there in dbpedia? > >> > >> > >> Regards, > >> Pooja > >> > > > > -- > | Rupert Westenthaler rupert.westentha...@gmail.com > | Bodenlehenstraße 11 ++43-699-11108907 > | A-5500 Bischofshofen > | REDLINK.CO > .......................................................................... > | http://redlink.co/ >
