Stanford CoreNLP
A Suite of Core NLP Tools
The coreference section of the output has the following description:
- The outer <coreference> tag signifies beginning of the whole
coreference graph
- Each inner <coreference> element describes one entity (set of
coreferent mentions) in the graph
- Each <mention> is a noun, pronoun, etc, that makes up part
of the graph
- A <mention> has the representative=true attribute only if
it is the full name of the entity. Only one mention has
representative=true. In this version, it will always be the first
one.
- Each mention has a <sentence> block telling you which
sentence it is in
- Each mention then has <start> and <end> blocks
telling you which words comprise it. This is word-level indexing, with
indexing starting from 1, and the end position numbering the word
after the end of the mention.
- Each mention also has a <head> block telling you which word
is the head word, if you consider it as dependencies