brendano / stanford_corenlp_pywrapper

151 stars 59 forks source link

Added the code for capturing the sentiment case #20

Closed duncanhodges closed 9 years ago

duncanhodges commented 9 years ago

I've not changed your build script in the repo, but works fine for me with jar library paths in build.sh set as appropriate.

brendano commented 9 years ago

really, their sentiment analyzer is token-level for everything? i noticed their examples always are about sentence-level or phrase-level sentiment. can you point me to documentation that confirms this is the correct way to use their sentiment analyzer?

duncanhodges commented 9 years ago

The analyser allocates sentiment scores at a token level in conjunction with the dependency tree - I was adding this to the output; i've rewritten the section to add the output for the whole tree if this is more consistent. I'll close this pull and open another.