The Role Of Naïve Bayes, SVM, And Decision Trees In Sentiment Analysis

Abstract

Sentiment analysis is a critical area of study within natural language processing that aims to systematically identify and categorize opinions expressed in text data. This paper evaluates the performance of three prominent machine learning algorithms, Naïve Bayes, Support Vector Machines (SVM), and Decision Trees in their ability to conduct sentiment analysis. Through empirical testing on datasets composed of online product reviews, we compare the accuracy, efficiency, and applicability of each algorithm. Our results indicate that SVMs provide the highest accuracy (85%), effectively managing high-dimensional data and complex linguistic structures. However, Naïve Bayes offers unparalleled speed, making it ideal for real-time applications, while Decision Trees excel in interpretability, despite their susceptibility to overfitting. The study highlights significant challenges, including sarcasm detection, contextual dependency, and data bias, suggesting future research directions such as enhanced contextual analysis and the development of multimodal sentiment analysis systems. This research contributes to the ongoing advancement of sentiment analysis technologies, providing insights that can aid in the selection of appropriate algorithms.