Optimized hybrid deep learning for cross-linguistic sentiment analysis: a novel approach

Abstract Sentiment analysis, the process of extracting and classifying opinions expressed in text, has gained significant traction in different fields, such as market research, customer feedback, and social media analysis. However, traditional sentiment analysis methods often struggle with multilingual data due to the complexities of language variations and differences in culture. These limitations can be addressed with hybrid deep learning approaches. This paper proposes an optimized hybrid deep learning model for multilingual sentiment analysis, leveraging the Grey Wolf Optimization (GWO) algorithm to enhance the model’s performance. The proposed approach combines convolutional neural networks (CNNs) and long short-term memory networks (LSTMs) to capture both local and contextual sentiment information effectively. Additionally, the approach utilizes language-specific word embedding techniques to address the semantic differences between languages. Unlike previous techniques, our model includes the Grey Wolf Optimization (GWO) algorithm for hyperparameter tweaking, which improves both accuracy and efficiency in multilingual sentiment analysis. It improves sentiment classification accuracy by 9.51% over typical machine learning models and 5.42% above deep learning baselines.