Unifying Sentiment Analysis and Emotion Recognition for Bangla Text: A Hybrid Approach

Text analysis refers to finding the inner meaning of textual data where it may vary based on the representation. Bangla text analysis is a trending topic due to its huge demand and hence researchers are devoted to finding new ways regularly to do so. The current works on Bangla text analysis either focused on Sentiment Analysis (SA) or Emotion Recognition (ER) separately. Moreover, the works utilized a single method like machine learning, deep learning, or rule-based approach solely. Hence, there is a lack of a framework that simultaneously conducts both SA and ER through a hybrid approach to enhance the analyzing performance. In this paper, we have proposed a framework to perform both SA and ER on Bangla text data through a newly developed hybrid approach. It initially applies five machine learning algorithms on two manually generated datasets (for SA and ER) separately and then picks the best outcome via the developed hybrid approach for both datasets. Finally, the experimental analysis shows the efficacy of our proposed framework over state-of-the-art works.