Sentiment Analysis Unveiled: Comparative Insights into Machine Learning Techniques Optimized by PSO and ACO

Abstract

Purpose: This paper contributes to sentiment analysis for customer reviews, focusing on analyzing records from a variety of tweets, which are often unstructured and can be positive, negative, or neutral.

Design/Methodology/Approach: To accomplish this, we started by organizing the data, extracting important adjectives as features, choosing how to represent these features, and using various machine learning algorithms like Naive Bayes, Maximum Entropy, and SVM. We also utilized semantic orientation based on WordNet to extract synonyms and similarities for textual features.

Findings: The study evaluates the classifier’s performance in terms of recall, precision, accuracy and F1-score.

Originality/Value: The paper’s value lies in its contribution to sentiment analysis for customer reviews, utilizing a variety of tweets and applying machine learning algorithms along with semantic orientation based on WordNet.

Paper Type: View Point