Sentiment Analysis on Web and Social Media Texts: A Comparative Study of Methods and Limitations

Authors: Erugu Krishna, Vijay Ramnath Sonawane

Full-text Research PDF File:   View   |   Download

Abstract: Sentiment analysis has emerged as a crucial research area for understanding opinions expressed in web and social media texts, driven by the exponential growth of user-generated content on platforms such as social networks, micro-blogs, forums, and e-commerce websites. These texts provide valuable insights into public perception, consumer behavior, and societal trends. However, analyzing such data is challenging due to informal language, domain dependency, multilingual diversity, ambiguity, and implicit sentiment expressions. This paper presents a comparative study of sentiment analysis methods applied to web and social media texts, focusing on lexicon-based approaches, traditional machine learning models, and advanced deep learning techniques. By systematically reviewing existing studies, we analyze the methodologies, datasets, performance outcomes, advantages, and limitations of each approach. The study highlights that while lexicon-based methods offer interpretability and efficiency, they suffer from limited coverage, whereas machine learning and deep learning models provide higher accuracy at the cost of data dependency and computational complexity. Comparative analysis reveals persistent research gaps related to dataset scarcity, cross-domain generalization, and standardized evaluation practices. This study aims to assist researchers in selecting appropriate sentiment analysis techniques and to identify promising directions for future research.

Keywords: Sentiment Analysis, Opinion Mining, Web Text, Social Media Analysis, Machine Learning, Deep Learning

Paper Id: 7.709

Published On: 2025-01-15

Published In: Special Issue - International Conference on Engineering, Economics, Management and Applied Sciences (January 2025)

Share this