Bengali Sentiment Analysis Using Acos Quad Classification: Integrating Machine Learning and Deep Learning Technique

Abstract

Aspect-Category-Opinion-Sentiment (ACOS) Quad classification represents a vital progression in Bengali sentiment analysis, addressing a notable gap in research for a language spoken by over 226 million people. While previous studies have explored sentiment analysis in Bengali, the ACOS framework remains underutilized. A comprehensive dataset has been curated and annotated with 13 distinct aspects across five categories to support this study. By leveraging a range of models—including BERT, BiLSTM, RNN, KNN, SVM, and T5—nuanced insights from usergenerated content are uncovered. This research enhances the understanding of sentiment in Bengali and demonstrates the effectiveness of transformer-based models in capturing intricate contextual nuances.

Keywords: spect-Category-Opinion-Sentiment Quad Ex-traction, ACOS, Bengali ACOS, Sentiment Analysis, BengaliSentiment Analysis