Analisis Sentimen Terhadap Video Ulasan Produk Menggunakan Metode Support Vector Machine Dengan Sequential Minimal Optimization

Abstract

The popularity of video as a medium for reviewing a product today has created interest and dependence on other users to look for video recommendations before buying the desired effect. Youtube social media is one of the media that has product reviews in the form of videos. The use of sentiment analysis can predict the tendency of someone's video review to have a positive or negative opinion which can be done by processing the video review into text form first using speech recognition. In this study, thoughts that have been in the form of text will be followed by the process of tokenizing, weighting and classification by applying the Support Vector Machine (SVM) algorithm model using Sequential Minimal Optimization (SMO) optimization. Based on the results of this study, it shows that the accuracy, recall, precision and f-measure values will increase with the increasing number of terms tested, while the C value variable in the enhanced SMO cannot be retrieved because the resulting accuracy value fluctuates. The test results with term 300 and C 1.5 get the highest value: accuracy 89.91%, recall 89.12%, precision 94.97% and f-measure 91.05.