I often run into the situation that I build a pipleine with pos tagger and
parser only to notice later that the parser also creates pos tags and that I
have everything twice then.
I believe we should disable the creation of pos tags in constituency parsers.
Also, we should make sure the parsers make use of existing POS tags if that is
technically possible. At least the Stanford and Berkeley parsers can do that
(not implemented in the BerkeleyParser DKPro component yet). Not sure about
other constituency parsers though.
Original issue reported on code.google.com by richard.eckart on 6 Aug 2014 at 7:08
Original issue reported on code.google.com by
richard.eckart
on 6 Aug 2014 at 7:08