Multilingual Sentiment Analysis Using the Social Eagle-Based Bidirectional Long Short-Term Memory

Abstract

Rapid evolvement in the field of social media constitutes an increasing demand in sentiment prediction for effective communication. Sentimental analyses are employed to extract opinions from individuals in order to invigorate the quality of material or a product for the effective growth of an organization. Researchers are aspired to predict the sentiment through comments of different languages, but there is a lack in the dictionaries accessible for different languages. In this research, hybridized Social Eagle Algorithm-based deep bidirectional long short-term memory (SoEo Algorithm-based deep BiLSTM) is proposed to predict the sentiment effectively. The languages are identified and converted into a standard format by the process of Transliteration and the features are extracted from these standardized data. The hunting strategy of the bald eagle and the adaptation behavior of coyotes are hybridized and executed in both forward and backward directions utilizing the BiLSTM classifier. The simulation outcome shows that the proposed model obtained an accuracy of 91.572%, precision of 89.196 %, recall of 91.551 % and F1 measure of 89.019%, which will be more efficient compared to the state-of-art methods