Deciphering Emotions: Unveiling Sentiment in Comments Through A-DNN-Based Analysis
S. S. Uma Sankari, S. Silvia Priscila · 2024
The task of conducting sentiment analysis on textual data poses significant challenges due to the intricate nature of language and the subjective aspects of basic human communication. Attention-based Deep Neural Network (A-DNN) is proposed as a solution for doing sentiment analysis on YouTube comments. The A-DNN model utilizes attention mechanisms and deep learning techniques to improve the accuracy of categorization. The extraction of exact semantic information is conducted by focusing on critical segments of the input text. To overcome the challenges of using low-quality training data and the need for cross-modal sentiment analysis, comprehensive preprocessing techniques are employed and attention mechanisms are included into the model's structure. The experimental results indicate that the A-DNN model demonstrates higher efficacy and robustness in comparison to other baseline models. The model's overall adaptability is enhanced by its structural flexibility and its ability to effectively handle various online sentiment analysis jobs. The model achieved an overall accuracy of 99.49% depicting the efficiency and reliability of the proposed model.