Artificial Intelligence Oriented User Sentiment Evaluation System on Social Networks using Modified Deep Learning Principles
S. Thangamayan, Sandip Nivrutti Jagdale, T. R. Kalai Lakshmi, Jayendra Gopal Thatipudi, K. Penyameen, Bhola Khan
Abstract
Sentiment analysis, a pivotal aspect of natural language processing, is vital for deciphering the emotional nuances embedded in textual data. This project introduces an innovative workflow that amalgamates various techniques to elevate sentiment analysis performance. The workflow encompasses Fuzzy C-means (FCM) clustering for pattern identification, Differential Multi-layer Whale Optimization (DWO) for feature extraction, and a classification phase utilizing diverse deep learning models, including Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN). Additionally, Ant Lion Optimization (ALO) ensures system optimization, fostering robustness and adaptability. The proposed system is evaluated against eight distinct deep learning models, each offering unique architectures or configurations. Remarkably, the proposed workflow achieves an impressive accuracy of 96%, underscoring its efficacy in discerning complex sentiments. This high level of accuracy positions the system as a powerful tool for real-world applications where nuanced understanding of human sentiment is paramount. The project contributes to the evolving landscape of sentiment analysis, providing a comprehensive framework for extracting valuable insights from diverse textual datasets.