Analyzing qualitative feedback using sentiment analysis / Muhammad Mazuan Mahmood

This paper aims to create a system to automated qualitative feedback analysis based on response in the survey. It is a promising technique to analyse the respondent textual responses by saving time and more efficiency than manual tally. This automated system can gain additional useful information to instructor performance evaluation from the respondent’s perspective. The survey data were collected from Office of Student Affair based on program organized by them. Using Naïve Bayes technique, the response ware tokenized to single word. Every words called lexicon have their own value. The value of the lexicons were taken from list of AFFIN words. These words have either positive or negative value. Naïve Bayes treat each response as a bag of word. The points for each words are added to get the total point for each response. To get the value of the words, it was compared to the list and take their value. The result was analysed based on the total point for each questions and for overall evaluation. The top five highest point are viewed based on its positivity or negativity.