Circle Chaotic Mapping Based White Shark Optimization Algorithm for Sentiment Analysis Classification

Sentiment Analysis (SA) helps consumers generate indistinct and detached sentiment recommendations based on huge volumes of information, which addresses the inherent human flaws in individual assessments. It aims to explore people's feelings, sentiments and attitudes through social media platforms such as blogs and forums, on topics related to issues, events, and entities, whose attributes are expressed through text assessments or reviews. SA faces challenges in accurately classifying text due to the ambiguity and contextual dependencies inherent in natural language. In this research, the Circle Chaotic Mapping based White Shark Optimization (CCM-WSO) is proposed for sentiment analysis. In traditional WSO, the CCM is incorporated to generate the initial population of white sharks, which enhances exploration capabilities, avoiding premature convergence and allowing the algorithm to efficiently search the solution space. Long Short-Term Memory (LSTM) is used to classify SA into categories such as positive, neutral and negative. The proposed CCM-WSO achieved an accuracy of 97.25%, outperforming other existing approaches such as THHO-Bi-LSTM, LSTM and RNN.