Paul Jansen Paul Jansen - 1 month ago 17
Java Question

Is there a POS tagger being able to parse "When did Nixon die?" correctly?

I am using the Stanford POS tagger to parse questions. It is one of the best POS taggers available in my experience, but a simple question like "When did Nixon die?" is incorrectly annotated as


When_WRB did_VBD Nixon_NNP die_NN


The problem is "die_NN". Obviously "die" is not a noun in this case but a verb. This happens with more questions. My questions are:


  • Is it possible to tweak the Stanford POS tagger to give the correct annotation for this sentence? Or alternatively, let it produce multiple annotations of which one is correct?

  • Are there any other POS taggers with a Java interface that are able to parse such questions correctly?


Answer

As stated in the comments, "The Stanford POS tagger currently gets this correct."

Lol... I'll take the bounty if he doesn't want it.