Analisis Sentimen Ulasan Aplikasi WeTV Untuk Peningkatan Layanan Menggunakan Metode K-Nearst Neighbor

Abstract

Online streaming applications are activities for watching movies that are very popular with the public, one of which is WeTV. WeTV is an online streaming that is used by the public as a medium of entertainment. The WeTV application has a rating of 4 out of 256 thousand reviews written by its users. The collection of reviews in the form of text can be collected and classified into negative responses, neutral responses, and negative responses. Positive responses are comments that are optimistic or supportive. Negative responses are comments on phrases or cases that do not support statements about related cases. Neutral responses are comments that are difficult to understand between negative or positive in nature to provide suggestions, sentences that have reviews from application users can be positive, negative and neutral, the data will go through a classification process using the K-Nearst Neighbor method. In this study, 12,000 reviews were used from the playstore. The research used the preprocessing stage, namely cleaning, case folding, tokenizing, normalization, stopword removal and steaming then to the TF-IDF stage and the final results will be tested with a confusion matrix using the Python programming language. The highest accuracy results from the testing process with a value of K = 3 in the dataset model 90% training data and 10% test data obtain an accuracy of 0.70%, precision 0.76%, recall 0.69%, f1-score 0.72% . Based on the results of the research that the K-Nearest Neighbor method is good in the process of identifying negative responses on WeTV.