How to solve ambiguity in sentiment analysis?

ikayunida123

Hello everyone :womanhappy:

I'm quite new to the text mining and I'm trying to do the sentiment analysis task today. But I encounter some problems while doing this sentiment analysis :womanfrustrated:

In my language, a word can have some different meanings. Like "setan" can have means : 1) devils 2) cursing words.

How to solve this ambiguity in sentiment analysis? Any tips would be great. Thank you!

Note : The algorthm that I'm using right now is Naive Bayes Classifier.

JEdward

In some cases the algorithm will actually solve the ambiguity for you. 

 

Naive Bayes works by looking at how often a word (or combination of words, if using n_grams) when put together have a certain meaning.  If your dataset shows that 60% of the time "setan" means devils & the remaining 40% means cursing then your scoring will mark the meaning as devils. 

N grams are where you combine separate words into a single one for example individually "setan" is 60% meaning devil, but "stupid_setan" is 90% of the time cursing.  

Naive Bayes is a nice algo that does not need to classify things in a binary way of true & false, but a range of possible choices.