1

I'm currently using Stanford NLP library for sentiment analysis of a twitter stream (version 3.3.0 but that’s not set.)

I was looking for ways to increase the accuracy when I came across this https://gate.ac.uk/wiki/twitter-postagger.html

I'm relatively new to sentiment analysis but am I right in saying that if I choose this model instead of the default model (which is based off film reviews) I would get an increased accuracy rating ?

If so, how does one go about integrating it with the Stanford NPL library ?

If I am missing any required information here please tell me !

Regards.

Richard
  • 13
  • 2
  • possible duplicate of [How do you use the GATE Twitter part-of-speech tagger as a model in the StanfordCoreNLP code?](http://stackoverflow.com/questions/20110712/how-do-you-use-the-gate-twitter-part-of-speech-tagger-as-a-model-in-the-stanford) – Jon Gauthier Apr 24 '15 at 03:42

2 Answers2

1

You can use GATE twitter pos model with stanford package

./corenlp.sh -file tweets.txt -pos.model gate-EN-twitter.model -ssplit.newlineIsSentenceBreak always

use v3.3.1 for GATE

user123
  • 5,269
  • 16
  • 73
  • 121
  • Thank you much appreciated , do you know if that would that result in a increased sentiment accuracy given the PoS would be more suited to tagging tweets ? – Richard Apr 24 '15 at 10:49
  • @Richard: as per I remember, stanford gives sentiment analysis for sentence level only, right? yes, I guess given a POS, accuracy may be increased. But if you write your own sentiment algo, there are more space to check various posibilities – user123 Apr 24 '15 at 10:54
  • Thanks, it's just the accuracy I'm getting is little better than random which I'm assuming is a result of the difference sentence structure and word usage found in tweets vs normal speech. – Richard Apr 24 '15 at 10:56
  • @Richard: you get sentiment result for entire tweet or what? I used stanford sentiment model long back, dont know if any change – user123 Apr 24 '15 at 10:58
  • Just the tweet message. – Richard Apr 24 '15 at 11:01
  • @Richard: I am much interested in sentiment part, would be glad to connect you if you are interested karimkhan.it@gmail.com – user123 Apr 24 '15 at 11:05
0

You can also have a look at the TwitIE pipeline: https://gate.ac.uk/wiki/twitie.html

This may probably enhance your tweet NER analysis and by the way your sentiment analysis.

Chester Mc Allister
  • 437
  • 3
  • 19