Comparison of Classification Algorithms on Product Boycott Sentiment Analysis in Real Time

Real-time monitoring of boycotted products is crucial for safeguarding brand reputation, maintaining customer trust, and enabling data-driven decision-making in today's rapidly changing consumer landscape. As social media platforms, particularly Twitter, become significant venues for public discourse, companies can harness sentiment analysis to capture and understand public opinion. However, sentiment analysis demands robust methods capable of processing large datasets while handling the complexity of long-term dependencies within the text. The effectiveness of real-time sentiment analysis relies heavily on selecting the most suitable algorithm. This paper aims to compare the performance of two prominent algorithms in text mining, Long Short-Term Memory (LSTM) and Naive Bayes, focusing on real-time sentiment analysis using boycott-related data from Twitter. Both algorithms were tested for their accuracy and efficiency in classifying sentiment data in a real-time setting. The results demonstrate that Naive Bayes consistently outperforms LSTM, achieving a remarkable accuracy rate of 95.23%. This superior performance highlights Naive Bayes as a more reliable and efficient option for real-time sentiment classification, especially in the context of boycotted products. These findings have provided valuable insights for businesses seeking to monitor consumer sentiments and make informed decisions in real-time, emphasizing the importance of algorithm choice in real-time sentiment analysis.