Scholarlink

This study explores sentiment analysis in Assamese, a low-resource language, using deep learning methods like CNNs and LSTMs with attention mechanisms. A balanced dataset of 2000 Assamese social media sentences was processed through cleaning, tokenization, and word embedding. Models incorporating attention layers were trained, with the LSTM + Attention model outperforming the CNN + Attention model, achieving 97.63% accuracy, 97.3% precision, 93% recall, and a 97% F1-score. The research addresses challenges like limited annotated data and linguistic complexities and suggests future directions, including cross-domain analysis, multilingual models, and practical applications for monitoring social media sentiments.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Attention-Based Deep Learning Approach for Sentiment Analysis in the Low-Resource Assamese Language

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Tulika Chutia,
  • HirokJyoti Saikia,
  • Sachin Deori,
  • Nomi Baruah

摘要

This study explores sentiment analysis in Assamese, a low-resource language, using deep learning methods like CNNs and LSTMs with attention mechanisms. A balanced dataset of 2000 Assamese social media sentences was processed through cleaning, tokenization, and word embedding. Models incorporating attention layers were trained, with the LSTM + Attention model outperforming the CNN + Attention model, achieving 97.63% accuracy, 97.3% precision, 93% recall, and a 97% F1-score. The research addresses challenges like limited annotated data and linguistic complexities and suggests future directions, including cross-domain analysis, multilingual models, and practical applications for monitoring social media sentiments.