An Automated System of Sentiment Analysis from Bangla Text using Supervised Learning Techniques

Abstract

Sentiment analysis has become a leading context for scientific and commercial market research in the field of machine learning. Currently, it's a more prominent research field of Bangla language processing system as there are few research works regarding sentiment analysis for this language. In essence, sentiment analysis is an automated process of text mining to determine the emotion from a given text. By using sentiment analysis, a given text can be categorized into several emotions. This paper deals with six individual emotion classes-happy, sad, tender, excited, angry and scared. Here, we proposed two methods of machine learning techniques- Naïve Bayes Classification Algorithm and Topical approach to extract the emotion from any Bangla text. Proposed methods have been applied for both article and sentence level of scope. A comparative analysis of the performance between these two methods has been done, and the topical approach achieved the best performance for both levels of magnitude.