Identifying Opinions of Footwear Products in Indonesia via Sentiment Analysis

Abstract

In the era of e-commerce has significantly altered consumer interactions, this study used the Naïve Bayes Classifier to analyze sentiments from a total 1,103 data collected from Shopee. Reviews were categorized into three sentiments to provide insights for potential buyers and manufacturers. The Naïve Bayes algorithm demonstrated an overall accuracy of 87.78 precent for the dataset and 95.59 percent for self-training data. Preprocessing steps included text cleaning, normalization, and tokenization to ensure data quality. Confusion matric for three different brands revealed effective sentiment classification. Future research could extend these findings by examining additional datasets or advanced machine learning techniques to further refine sentiment classification approaches.