4

I am a computer scientist using Stanford NLP for extracting a Semantic Graph from plain text.
Through this tool I am already getting the universal dependencies but now I want to get all the possible types of indirect objects through the nmod relation.

In Italian we distinguish if the indirect complement is either motion towards a place (e.g. I am going to Paris), a “term” complement (e.g. I gave the book to Mary), placement (e.g. It was held in Rome), time (e.g. It was held in 1999) motion from a place (e.g. I am coming from Switzerland), mean complement (e.g. travelling by bike), Agent Object (e.g. It was caused by a tornado).

Is there an automatic tool through which I can distinguish such elements programmatically (e.g. via Java / Stanford NLP) by using only the verb alongside with the preposition and some information from the object? Thanks in advance.

4
  • Those are not indirect objects but obliques. Generally such information is if semantic nature so you can't distinguish most of them without a semantic lexicon.
    – Atamiri
    Aug 13 '16 at 18:46
  • 1
    Ok, that's because in Italian we call them “indiretti” that could be literally translated as “indirect”, thanks. Is there a way to acheive such semantic lexicon?
    – jackb
    Aug 13 '16 at 18:52
  • 1
    You might be interested in the English version of Vallex, it's a valency lexicon with mappings from syntactic relations to semantic roles (of what the authors call "free modifiers" which roughly correspond to obliques).
    – Atamiri
    Aug 13 '16 at 19:29
  • So ok, the only thing is to use something like Vallex or/and Babelnet (where I have the same thing, but without corpora).
    – jackb
    Aug 13 '16 at 20:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.