Machine Learning Algorithms for Sentiment Classification: Comparing Accuracy of SVM, Random Forest, and LSTM

EasyChair Preprint 14882

Abstract

Sentiment classification, a vital task in natural language processing, seeks to determine the sentiment behind textual data, such as customer reviews or social media posts. This paper compares the performance of three widely used machine learning algorithms for sentiment analysis: Support Vector Machine (SVM), Random Forest (RF), and Long Short-Term Memory (LSTM). SVM and RF, traditional machine learning methods, excel at classifying structured, non-sequential data, while LSTM, a type of recurrent neural network, is designed to capture the sequential dependencies in text. We evaluate these algorithms based on their accuracy, computational efficiency, and ability to handle complex language structures across different datasets. Our results demonstrate that while SVM and Random Forest perform adequately on smaller datasets with simpler features, LSTM significantly outperforms them in capturing nuanced contextual information, albeit at a higher computational cost. This study provides insights into the trade-offs between traditional and deep learning approaches, offering guidance on algorithm selection for sentiment classification tasks.

Keyphrases: Long Short-Term Memory, Random Forest, Support Vector Machine, sentiment classification

@booklet{EasyChair:14882,
  author    = {Docas Akinyele and Godwin Olaoye and David Ray},
  title     = {Machine Learning Algorithms for Sentiment Classification: Comparing Accuracy of SVM, Random Forest, and LSTM},
  howpublished = {EasyChair Preprint 14882},
  year      = {EasyChair, 2024}}