I plan to work with dependency parsing, as an intermediate step in a text understanding problem. I'm also still trying to understand the terminology and methods. Any pointers (or providers) of tools (algorithms, python code, toy-this or that) so that I can implement a parser for purposes of learning about the mechanics of dependency parsing?

I understand that implementing a real parser is a major task. I'm hopeful for just enough information to code a 'toy', to learn more. Yes, I'm hoping I could insert some intermediate "print" statements to dump the state of the parser.

Really simple sentences like "Farmers grow wheat" to start with, then adding a few additional bits "Farmers grow wheat in Ohio", and "Farmers grow wheat in Ohio with water" (which has a few issues to be resolved for understanding).

  • For constituency parsing I would have advised pset 2 of Michael Collins' NLP MOOC: class.coursera.org/nlangp-001/assignment/view?assignment_id=5 Sep 19, 2015 at 23:52
  • I will post Python code that implements a relatively simple dependency parser if that's helpful (implemented this in my college years, back when the dinosaurs roamed the earth). Sep 21, 2015 at 3:00
  • Thank you. I will probably learn enough to "rediscover" that I should use an existing parser. That's ok. Finding out what I don't know is all part of the process! Your response appreciated !
    – Anon
    Sep 21, 2015 at 3:28
  • You could use Answer Set Programming, there's an example on Wikipedia: en.wikipedia.org/wiki/Answer_set_programming#Dependency_parsing
    – Atamiri
    Sep 22, 2015 at 13:08
  • 1
    Please don't edit the body of your question so that it drastically changes its content.
    – Alenanno
    Sep 26, 2015 at 10:43

2 Answers 2


spaCy.io has a nice high-accuracy lightweight parser.

If you only need English then it is a good choice.


Good question. Try NLTK's dependency parser. Seems to work. Watch for noise in this forum. It is over- moderatoando.NP meaning get your boots ready. Now Ricky has some 'splainibg to do.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.