Sentiment Analysis for Low Resource Language Using Machine Learning and Deep Learning

Abstract

The challenging task of sentiment analysis on Punjabi and other languages with limited resources and technology is investigated in this work. Though sentiment analysis is a prominent topic in NLP, low-resource languages like Punjabi lack digital resources, annotated datasets, and language-specific tools, so research in them is ignored. This paper addresses such information gaps by means of Punjabi text sentiment analysis. Internet news and social media posts were compiled into an annotated Punjabi dataset with sentiment polarity. Text sentiment was extracted and analysed using deep learning tools and rule-based machine learning methods—hybrid models and supervised classifiers. We preprocessed carefully to manage Punjabi's transliterations, dialectal variances, and cultural terms. We evaluated model performance with reference to recall, accuracy, precision, and F1-score. Despite few resources, deep learning models identified more complex sentiment signals than machine learning methods. For slang, rule-based sentiment analysis also performed admirably. We draw attention to problems with Punjabi sentiment analysis including a lack of high-quality annotated data and a platform-wide shared script. While advanced natural language processing techniques might be useful, more data and research will let you create better models for low-resource languages. This work gives a Punjabi sentiment analysis method, a platform for further low-resource language processing research, and defines the cultural background required for correct sentiment interpretation.