Sentiment Prediction for Social Information Retrieval: A Comparative Study of Machine Learning and Deep Learning Approaches
Aicha Boubekeur, Fouzia Benchikha, Naila Marir · Advances in intelligent systems research/Advances in Intelligent Systems Research · 2024
Sentiment analysis plays a pivotal role in social information retrieval, enabling the extraction of valuable insights from user-generated content.In this study, we conduct a comprehensive comparative analysis of machine learning and deep learning approaches for sentiment prediction in the context of social media data, with a specific focus on the COVID-19 vaccine discourse.We investigate the performance of traditional machine learning classifiers, including Naive Bayes, Support Vector Machines, K-Nearest Neighbors, and Decision Tree, in conjunction with the TF-IDF representation model.In parallel, we assess the efficacy of deep learning models, such as Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and a hybrid LSTM-CNN architecture, utilizing Word Embedding representation.Notably, the CNN model with Word2Vec vectorization demonstrates the highest performance.While the accuracy of the combined model, featuring the two LSTM-CNN classifiers, is slightly lower for our specific problem.