Eksplorasi Algoritme Klasifikasi untuk Analisis Sentimen pada Ulasan Google Review Terhadap Ekowisata Sungai Mudal

Mudal River is an ecotourism developed by PT. PLN (Persero) located in Kulon Progo Regency, Special Region of Yogyakarta. As a tourism developer, visitors' impressions can be used as evaluation material to improve tourism facilities and services. This study aims to determine the distribution of positive and negative sentiments from visitor reviews that can be used as support in business decision making. A total of 3,663 data were obtained from Google Review, then pre-processing was carried out to obtain higher quality data. Labeling/annotation uses Large Language Models (chatGPT), while the classification process uses the Stochastic Gradient Descent Method. Model testing uses the K-Fold method to obtain average accuracy, recall, and AUC values. The classification performance results produce an accuracy of 85.5%, with a dominance of positive labels in the annotation results. Meanwhile, the wordcloud results show that words such as water quality, waterfalls, mudal river, and photo spots have positive affirmations. However, several areas are of concern in negative sentiment, especially related to roads and photo spot management, as well as the issue of entrance tickets and the potential for paid photos.