TUNICATE SWARM OPTIMIZATION WITH DEEP LEARNING MODEL FOR NLP BASED SENTIMENT ANALYSIS AND CLASSIFICATION MODEL

Abstract

By the online occurrence of majority of the world population, social media roles a most essential play from the lives of individuals and businesses similar. The social media allows businesses for advertising its product, form brand values, and obtained its customers. For leveraging these social media platform, it can be essential to businesses for proceeding customer feedback from the procedure of tweets and posts. Sentiment analysis (SA) is the procedure of recognizing the emotion, both negative, positive, and neutral, connected to these social media texts. This paper focuses on the design of Tunicate swarm optimization with deep learning model for SA and classification (TSODL-SAC) model. The presented TSODL-SAC model focuses on the identification and categorization of distinct sentiment classes. To attain this, the presented TSODL-SAC model applies data pre-processing to transform the data into useful data. In addition, the TSODL-SAC model employs word2vector model to derive feature vectors. Moreover, the TSO algorithm with long short term memory (LSTM) model was utilized for categorizing different classes of sentiments. Furthermore, the TSO system was exploited for optimum tune the hyperparameter related to the LSTM approach. The experimental validation of the TSODL-SAC algorithm was applied to benchmark dataset and the outcomes were inspected with respect to various measures.