Replace NPs in parse tree with coreference ID

charmainekeck / PyCoref

Final Project for CS 5340 / 6340, Coreference Resolution

http://www.eng.utah.edu/~cs5340/project.html

3 stars 0 forks source link

Replace NPs in parse tree with coreference ID #2

Open adamwalz opened 12 years ago

adamwalz commented 12 years ago

Parse trees of the form

          S
      /        \
     NP        VP 
    /  \        |
   the dog     ran

Should be replaced with

          S
      /        \
     $1        VP 
    /  \        |
   the dog     ran

Where there is a coreference tag <COREF ID="1">the dog</COREF> in the text xml

adamwalz commented 11 years ago

Incorrect attachement may be fixed by looping through trees and corefs in reverse order. To do this however, corefs need to be stored as an ordered list, not a dictionary. Created issue #8 for this.

adamwalz commented 11 years ago

Having about 93% of corefs tagged in trees will allow us to move on with resolutions (Hobbs algorithm). The rest of this issue can move into a different iteration