Abstract:

With the rapid growth of digital transformation, social media and online platforms have become essential spaces for individuals to express their opinions and experiences. Analyzing this user-generated content is crucial for organizations to understand customer sentiments and improve decision-making. Sentiment analysis is a key technique in Natural Language Processing (NLP) that interprets emotions from textual data, categorizing them as positive or negative. This study explores the preprocessing and data preparation steps involved in sentiment analysis and presents a comparative evaluation of various classification models. We assess the performance of traditional machine learning techniques, including Support Vector Machine (SVM) and Multinomial Naïve Bayes (MNB), alongside deep learning methods such as Bidirectional Encoder Representations from Transformers (BERT) and Long Short-Term Memory (LSTM). Our findings highlight the strengths and limitations of these approaches in sentiment classification.

Description:

.