Changes for version 0.03
- IDs are now sequential, rather than random. (Non-deterministic IDs are no fun after a a failure in the middle of processing 11,000 files (and God knows how many sentences).)
- Smarter getting of index for governor/dependent in dependencies. (IndexedWord has an, wait for it, index method. I should read javadocs more closely)
- Dependencies now have full names (e.g. 'prep_of' rather than 'prep'). Thanks to Eneko Agirre for the fix.
- Lingua::StanfordCoreNLP - A Perl interface to Stanford's CoreNLP tool set.