New BERT-CNN-BiLSTM model advances Bangla news headline classification and sentiment analysis

Researchers have developed a novel BERT-CNN-BiLSTM framework designed to simultaneously classify Bangla news headlines and analyze their sentiment. This hybrid transfer learning model was tested on the BAN-ABSA dataset, comprising over 9,000 headlines, and demonstrated superior performance compared to baseline models. The study explored two experimental strategies for handling imbalanced data, with one approach yielding headline and sentiment classification accuracies of 81.37% and 64.46%, respectively, establishing a new state-of-the-art for Bangla text classification in low-resource settings. AI

IMPACT Establishes a new baseline for Bangla text classification, potentially improving information access in low-resource language environments.

RANK_REASON Academic paper detailing a new NLP model and dataset. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• BAN-ABSA
• Bangla
• BERT-CNN-BiLSTM
• Saydul Akbar Murad

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →