I wanted to perform Twitter sentiment analysis on Twitter tweets. When I googled it, I found a [*stanfordNLP*][1] code of sentiment class. I used it for finding Twitter sentiment later I came to know [*TwitterGate*][2] model is made for Twitter tweets only. 

Then I replaced the *stanfordNLP* model with *TwitterGate* model. I did not find any difference in the sentiment results. Both models give the same result.
 
However, when I tested the same tweets with [*Monkeylearn*][3] API and [*Datumbox*][4] API sentiment, I got confused with results. In NLP and twitterGate most tweets sentiment is given as negative, where the same tweets sentiment is given as neutral in *Monkeylearn* & *Datumbox* APIs code. 

How can I know which model is giving the correct sentiment? 


  [1]: http://nlp.stanford.edu/
  [2]: https://gate.ac.uk/wiki/twitter-postagger.html
  [3]: http://monkeylearn.com/
  [4]: http://www.datumbox.com/