Extracting the relationship between entities in Stanford CoreNLP

Question

I want to extract the complete relationship between two entities using Stanford CoreNLP (or maybe other tools).

For example:

Windows is more popular than Linux.

This tool requires Java.

Football is the most popular game in the World.

What is the fastest way? And what is the best practice for that?

Thanks in advance

Answer 1

ReVerb focuses on Open IE. You can start by reading their paper "Identifying Relations for Open Information Extraction" and checking the demo site.

Answer 2

You are probably looking for dependency relations between nouns. Stanford Parser provides such output. Have a look here. You can combine what Pete said (i.e. the POS graph) with the dependency graph to identify what relationship (for example, direct object or nominal subject, etc.) a pair of nouns (or noun phrases) share.