Applied Linguistics With Red-Tailed Hawk Optimizer-Based Ensemble Learning Strategy in Natural Language Processing

• Published in: IEEE access 2023, Vol.11, p.132448-132456
• Main Authors: Alshahrani, Hala J., Hassan, Abdulkhaleq Q. A., Almalki, Nabil Sharaf, Alnfiai, Mrim M., Salama, Ahmed S., Hamza, Manar Ahmed
• Format: Article
• Language: English
• Subjects: Algorithms; Applied linguistics; Artificial neural networks; Classification; Convolutional neural networks; Data mining; Deep learning; Ensemble learning; Feature extraction; Learning strategies; Linguistics; Machine learning; Mass media; Natural language processing; red-tailed hawk optimizer; Sentiment analysis; Short term memory; Social networking (online); Social networks; Strategy; Task analysis; Tuning; Unstructured data
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2023.3331061

Natural Language Processing (NLP) is the most vital technology in currently utilized, specifically caused by the huge and growing count of online texts that requires that understood for its massive value that completely asserted. NLP is create a sense of unstructured data, which are created by social networks and other social data sources, and is supported to organize them into an additional structured model that assists several kinds of tasks and applications. Sentiment analysis (SA), a subfield of NLP contains determining the sentiment expressed or emotional tone from a piece of text. Deep learning (DL) approaches are significantly advanced the field of SA, permitting for more accurate and nuanced classification of sentiments from the text data. In this article, we present an Advanced Sentiment Analysis using a Red-Tailed Hawk Optimizer with Ensemble Learning (ASA-RTHEL) Strategy in NLP. The aim of ASA-RTHEL technique is to exploit the strategies of ensemble learning with a hyperparameter tuning process for SA. The ASA-RTHEL technique mainly follows an ensemble learning-based classification process, which combines prediction from three DL approaches convolutional neural network (CNN), gated recurrent unit (GRU), and long short-term memory (LSTM). The ensemble process results in enhanced SA performance and decreases the risk of depending only on a single model bias or error. To boost the SA performance, the hyperparameter tuning strategy is performed by the use of the RTH algorithm. An extensive set of experiments were carried out for ensuring the superior SA results of ASA-RTHEL technique. The comprehensive comparison study highlighted the enhanced results of the MPONLP-TSA method on the recognition of various kinds of sentiments.