A Machine Learning and DL Approach to Marathi Sentiment Analysis Using Senticnet

In the era of digital communication, understanding public sentiment is crucial. However, sentiment analysis tools for less common languages like Marathi are limited. This paper introduces a machine learning and deep learning approach to Marathi sentiment analysis using Senticnet. Through the utilization of Senticnet and various machine learning techniques, we collected and pre-processed data, adapt Senticnet for Marathi, and design language-specific sentiment analysis models. Leveraging techniques such as tokenization, text cleaning, and feature extraction, we effectively classified Marathi text into positive, negative, and neutral sentiments. Our library's performance is evaluated against existing tools, show casing its accuracy and sensitivity to Marathi sentiment nuances. This work not only enhances Marathi sentiment analysis but also offers insights into adapting resources for non-english languages. By sharing our methodology and library, we encourage further research in regional languages, promoting sentiment analysis in diverse linguistic landscapes. The significant this research focuses on the creation of robust Marathi sentiment analysis tool, facilitating deeper understanding and analysis of sentiments in underrepresented languages, and serving as a catalyst for future advancements in sentiment analysis across linguistic boundaries.