I am using IBM Watson to get sentiments of social media texts. But many of these texts are just emojis. Currently, I am not able to get any sentiment for emojis. I get unsupported text language error. Is there anyway to get the sentiments for emojis using Watson NLU?
{
"language": "unknown",
"error": "unsupported text language: unknown",
"code": 400
}
Provided you are sending them in as utf characters in a utf string, then it really depends on whether the NLU corpus has been trained to recognise them - I don't know if it is, however ...
I tried testing the following strings for document sentiment:
“I am feeling today” - Sentiment : -0.823 : Negative
“I am feeling today" - Sentiment : +0.939 : Positive
You need some text for the service to recognise the language and hence which corpus to use.
