Sentiment Analysis with Text Mining: A Study from the Newspaper Contents of Bangladesh

Analyzing public sentiment through news media is essential for understanding societal trends and opinions.This paper examines sentiment patterns in newspaper data from The Daily Sun and Dhaka Tribune, employing sentiment analysis techniques via web scraping.Despite the growing reliance on sentiment analysis, challenges remain in accurately capturing sentiment dynamics from diverse data sources, particularly news media.This paper addresses these challenges using emotion plot visualization, k-means clustering, and classifiers like Random Forest and Naive Bayes.Data was scraped from both newspapers, preprocessed, and analyzed to identify common themes and sentiment patterns.Techniques included parser selection for effective web scraping, clustering to group articles by sentiment, and applying machine learning classifiers to determine sentiment distribution.The analysis revealed that most articles in both newspapers exhibited a neutral sentiment throughout the year.The sentiment classifiers demonstrated high accuracy, with Random Forest achieving approximately 96% and Naive Bayes reaching 98%.The insights from this research can enhance predictive capabilities in sentiment analysis models and offer a deeper understanding of public opinion dynamics.Future research could explore integrating additional data sources and employing advanced models to further refine sentiment analysis accuracy.