Analisis Ulasan Pembelian Produk Beras di Marketplace Shopee dengan Menggunakan Topic Modelling

Abstract

This research is motivated by the motivation to capture customer experience in purchasing rice products in the Shopee marketplace. The method used is Latent Dirichlet Allocation and analyzed based on several rating categories, namely low rating (negative sentiment), medium rating (neutral sentiment) and high rating (positive sentiment). The results of this study indicate that customer reviews of purchasing rice products in the shopee marketplace have a dominant positive sentiment, followed by negative and neutral sentiment. Negative sentiment consists of the words with the most frequency namely "quality", "delivery", "price" as well as negative words such as ticks and leaks. Positive sentiment consists of the words that appear the most, namely "price", "buy", "quality", "send", "good". The selection results based on the coherence value are visualized with a total of 3 topics, namely those related to the topics of rice quality and price, taste and quality, price and delivery.