Hello,
If I increased the number of keywords (Features) of examples in training set pertaining to a particular class regarding to Sentiment Analysis Classification , will increase the efficiency of the classifier?, such as take the word "Happy" or "Hate" and all their Synonyms, so that it can independently classify the new sentences for any class belonging? To be noted, I deal with a large data set.
Many thanks in advance to your answers
Getting more features/information from certain class increase efficiency, however, if these features become redundant then it moved into more complex process. However, in this case, use histogram of BoW and chose the most important features and eliminate the redundant features.
Thanks.
Your question itself is not clear. What is your research problem and what is the data that you have, to start with?
I am not sure it would generally improve classification results. I would rather unify the features and extract positive feature cluster for "Happy" or negative feature cluster for "Hate" . You would have the same for negative feature "sad" and positive feature for "joy". You might be interested in my Thesis Opinion Mining and Lexical Affect Sensing
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training