Applied Intelligence and Computing
A Literature Review on Sentiment Analysis
Authors: Ahongsangbam Dorendro and Haobam Mamata Devi
Publishing Date: 02-01-2025
ISBN: 978-81-955020-9-7
Abstract
Sentiment analysis, or opinion mining, is a vital subfield of natural language processing (NLP) focused on identifying and classifying subjective information in text data. This review explores the evolution of sentiment analysis from early lexicon-based methods to contemporary deep learning techniques. Initially reliant on static word lists, early approaches, such as those introduced by Turney (2002), provided basic sentiment scoring but struggled with context and nuance. The shift to machine learning brought more dynamic models, like Support Vector Machines (SVMs) and Naive Bayes classifiers, which improved adaptability and accuracy. The deep learning revolution, highlighted by advancements such as Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, enabled more sophisticated sentiment understanding through feature extraction and sequential data processing. Transformer-based models, including BERT and GPT-4, further enhanced sentiment analysis by leveraging self-attention mechanisms and bidirectional context. Despite these advancements, challenges remain, such as handling sarcasm, domain-specific language, and multilingual analysis. Future directions include integrating multimodal data and improving model explainability, promising to advance the accuracy and applicability of sentiment analysis in various domains.
Keywords
Sentiment Analysis, Natural Language Processing (NLP), Deep Learning, Machine Learning, Transformer Models, Contextual Understanding, Sarcasm Detection.
Cite as
Ahongsangbam Dorendro and Haobam Mamata Devi, "A Literature Review on Sentiment Analysis", In: Mukesh Saraswat and Rajani Kumari (eds), Applied Intelligence and Computing, SCRS, India, 2025, pp. 303-312. https://doi.org/10.56155/978-81-955020-9-7-29