Peringkasan Review pada Situs Belanja Online Menggunakan NLP (Natural Language Processing) Berbasis Text Summarization

Abstract

Online marketplace or e-commerce market is a place where sellers and buyers make buying and selling transactions electronically because it has many conveniences. The growth of the e-commerce market in Indonesia continues to increase every year. Before making a purchase, buyers usually look at reviews on online shopping sites and compare some of the differences from these online shopping sites. However, some reviews use long statements so that buyers need to read carefully to find the meaning of the review. Therefore, it is necessary to do a text summarization technique and classify the reviews into positive or negative. This study uses data that is crawled from the Google Play Store website which is then pre-processed the data, namely case folding, stopwords removal, normalization, stemming, and tokenizing. In this research, a sequence-to-sequence model algorithm is used to summarize and combine Convolutional Neural Network (CNN) & Long Short-Term Memory in conducting sentiment analysis. Based on the tests conducted on buyer reviews, this study obtained an accuracy value of 0.87 Precision, 0.92 Recall and F-Measure 0.89 for summary and 90% accuracy value for sentiment analysis.