05-08-2017 12:09 AM - edited 05-08-2017 12:10 AM
We are using CDH 5. we have implemented twitter data sentiment analysis.
We have splitted the text into words and calcutating polarity. Sentiment It is working fine for few tweets.
and it is not working for few tweets..
for example : one tweet text is "not working as expected". here actual sentiment is Negative. but we are getting nuetral. Beacuse " not (-1) + working (1) + as (0) + expected (0) = 0".
My query is, can we calculate sentiment by phrases, instead of words ?
Thanks in advance.