search

cjhutto / vaderSentiment

VADER Sentiment Analysis. VADER (Valence Aware Dictionary and sEntiment Reasoner) is a lexicon and rule-based sentiment analysis tool that is specifically attuned to sentiments expressed in social media, and works well on texts from other domains.
MIT License
4.43k stars 1k forks source link

Not predicting sentiment of emoticons correctly #56

Closed Rishav09 closed 6 years ago

Rishav09 commented 6 years ago

It is not predicting inconsistent results on emoticons.For instance, when I am passing this as '🙂' an argument, it is correctly predicting the outcome but on using same emoticons multiple times '🙂🙂', it is giving neutral results.Similarly ,the same issue is arising in different cases of other emoji and sometimes ,it is not even detecting the single emoji too.

ayushbits commented 6 years ago

@cjhutto I am also facing the issue. Could you tell why is it happening ?

ckw017 commented 6 years ago

The analyzer splits up the input by whitespace and looks for emoji among those tokens. So when it encounters a token like two smiling faces in a row, it will look for the token containing both faces in the emoji lexicon, fail to find it, and carry on. A fix for this would be to search for emoji on a character by character basis.

ddugovic commented 6 years ago

Does :smiley: :smiley: mean the same thing as :smiley: ?

ckw017 commented 6 years ago

I feel like it should mean the same thing as if ":)" would be to ":) :)"

Rishav09 commented 6 years ago

@ddugovic No, it doesn't. 😃 😃 have more positive compound score than 😃 . @ckw017 Yes, I realised, emoji's needs to be space separated for analyser to work.