1
$\begingroup$

I'm speculating that this is very much known in NLP, but as I've not studied NLP, I don't know what concepts are related to this:

enter image description here

Basically I have entities that are connected by some strings. I want to do semantic analysis using this kind of constructs. I.e. by checking whether a given sentence has entities (the yellow ellipses) connected with the right strings.


Notice that this is also basically describable as a a deterministic context-free language. But I'd like to understand the semantical aspect of this, rather than mere parsing (string matching).

$\endgroup$
1
1
$\begingroup$

Given your example, Lexical Semantics may be what you are looking for. Things like WordNet provide hypernym (breakfast -> meal), hyponym (meal -> lunch), meronym (course -> meal), holonym (water -> oxygen) relationships. See chapter 19 of Speech and Language Processing by Jurafsky and Martin for more thorough set of details.

$\endgroup$
2
  • $\begingroup$ Any idea how are these implemented? SQL tables of something? $\endgroup$ – mavavilj Feb 9 '16 at 14:42
  • $\begingroup$ @mavavilj Use wn to find the desired types of relationships in the WordNet dataset. wn will provide formatted output that you can easily parse and store in the data structure of your choosing. You'd have to dig into the code to see how it's implemented. $\endgroup$ – GEL Feb 10 '16 at 4:12

Your Answer

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

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