INTEGRATING FRACTAL SNOW ABLATION OPTIMIZER WITH BAYESIAN MACHINE LEARNING FOR ASPECT-LEVEL SENTIMENT ANALYSIS ON SOCIAL MEDIA

Social media platforms have become vast repositories of user-generated content, offering an abundant data source for sentiment analysis (SA). SA is a natural language processing (NLP) algorithm that defines the sentiment or emotional tone expressed in the given text. It includes utilizing computational techniques to automatically detect and categorize the sentiment as negative, positive, or neutral. Aspect-based SA (ABSA) systems leverage machine learning (ML) approaches to discriminate nuanced opinions within the text, which break down sentiment through particular attributes or aspects of the subject matter. Businesses and researchers can gain deep insights into brand perception, public opinion, and product feedback by integrating social media data with ABSA methodologies. This enables the extraction of sentiment polarity and more actionable and targeted insights. By applying ML approaches trained on the abundance of social media data, organizations can identify areas for improvement, tailor their strategies to meet their audience’s evolving needs and preferences and better understand customer sentiments. In this view, this study develops a new Fractal Snow Ablation Optimizer with Bayesian Machine Learning for Aspect-Level Sentiment Analysis (SAOBML-ALSA) technique on social media. The SAOBML-ALSA approach examines social media content to identify sentiments into distinct classes. In the primary stage, the SAOBML-ALSA technique preprocesses the input social media content to transform it into a meaningful format. This is followed by a LeBERT-based word embedding process. The SAOBML-ALSA technique applies a Naïve Bayes (NB) classifier for ALSA. Eventually, the parameter selection of the NB classifier will be done using the SAO technique. The performance evaluation of the SAOBML-ALSA methodology was examined under the benchmark database. The experimental results stated that the SAOBML-ALSA technique exhibits promising performance compared to other models.