Group Abstract Group Abstract

Message Boards Message Boards

[WSC18] Using Twitter Sentiment Analysis to determine emoji sentiment

Posted 7 years ago
Attachments:
POSTED BY: Jake Frances
4 Replies

Yes. I also tried in Spanish to no avail. So a relatively easy road to improvement (given that as you say Mathematica recognizes languages very well) would we to allow sentiment analysis in different languages

Great contribution! I have a question here: does Classify["Sentiment", #] work with languages other than English? And if so then how? Thanks!

It doesn't seem to. For example, this very positive sentiment in German is classified as 'neutral':

 Classify["Sentiment", "Ich bin sehr, sehr glücklich. Mir geht es ausgesprochen gut."]

Playing around with a few more examples in German suggests the same. It's a pity, because Classify already does a nice job identifying languages. So it has the means to warn you if the target language is unsupported for certain classification tasks. I couldn't see language support addressed in the documentation either.

POSTED BY: Arno Bosse

Nice work Jake! You might be interested in extending this study by including your findings with emojis: "Diurnal variations of psychometric indicators in Twitter content"

Some functions that might be useful for such study are: TextStructure , PartOfSpeech

Reply to this discussion
Community posts can be styled and formatted using the Markdown syntax.
Reply Preview
Attachments
Remove
or Discard