Implementation of Aspect-Based Sentiment Analysis for Outdoor Gear Product Reviews on the Tokopedia E-Commerce Platform (Case Study: Eiger Adventure)

Abstract

This study aims to uncover customer perceptions of Eiger Adventure products on the Tokopedia platform through an Aspect-Based Sentiment Analysis (ABSA) approach. Data in the form of 3,670 customer reviews were collected using web scraping techniques and processed through preprocessing, aspect extraction, sentiment labeling, and classification using the K-Nearest Neighbor (KNN) algorithm. The text vectorization process was carried out using the TF-IDF method, while evaluation metrics included accuracy, precision, recall, and F1-score. The results showed that the most dominant aspects were quality, model, and delivery. Positive sentiment was mostly found in the quality and model aspects, while delivery and size aspects were the main sources of complaints. The classification model achieved an accuracy of 61.6% and performed best in detecting positive sentiment. As an implementation, a web-based system was developed that is capable of conducting ABSA analysis in an integrated and interactive manner. This study concluded that although Eiger products are considered superior in quality, efforts to improve non-product services such as delivery and size information are urgently needed to improve overall customer satisfaction.