Abstract:
Nowadays, accessing to video contents and services through social media such as YouTube is fast and easy. There are both positive and negative comments over the media, which influence consumer purchasing decisions and business information management. Therefore, this research focuses on the sentiment analysis from Thai videos on beauty reviews using data mining techniques. By collecting feedback on the beauty products from YouTube and applied with automatic speech recognition, word segmentation, comparison and decision, word adjustment, development of sentiment analysis systems from Thai reviews videos, evaluation of accuracy and performance evaluation of system satisfaction, document representation, sentiment analysis modeling and comparison of machine learning technique performance. The results of this study showed that word adjustment step was accurate at 92%. The performance of Machine Learning Technique for replacing word with term frequency (TF) with word adjustment step and applied with SVM technique and LibSVM function was accurate at 94.4% when K equals 10 from K-Fold Cross Validation