I went through their github files as well as the official site, I can't find the named entity tagging training corpus they used in splotlight.
How Can I found the dataset instead of a trained model?
Asked
Active
Viewed 1,170 times
3
-
Have you tried to check the repo https://github.com/dbpedia-spotlight/model-quickstarter ? – Sandro Athaide Dec 02 '14 at 22:08
-
I found the exact guide on ner dataset here: https://github.com/dbpedia-spotlight/pignlproc. Thanks a lot! – Tilney Dec 03 '14 at 01:32
-
Please post that as an answer and accept it so that this question no longer comes up as unanswered. Thanks. – tripleee Dec 05 '14 at 08:03
1 Answers
0
see This link https://github.com/dbpedia-spotlight/dbpedia-spotlight/wiki/Web-service
In here, method for setting up dbpedia lookup offline is explained. Also they have given 4 tar files which are
- redirects_en.nt
- short_abstracts_en.nt
- instance_types_en.nt
- article_categories_en.nt
these are supposed to be training data for it.
Gunjan
- 2,775
- 27
- 30
-
The link you refer to provides guide on using dbpedia-spotlight services, I didn't find any information on how to generate NER training corpus. It's true we can use the 4 tar files to generate ourselves, but the whole parsing process is time consuming and more importantly, it's not part of our core logic. So I was looking forward to a tool to generate ner training data as I posted before(http://github.com/dbpedia-spotlight/pignlproc) – Tilney Dec 15 '14 at 03:24