INTENT CLASSIFICATION OF USER COMMENTS IN MYANMAR LANGUAGE ON SOCIAL MEDIA SHOPPING PAGES

Abstract

According to the social media usage statistics, there are about four billion total social media users across all platform. Moreover, people spend more time on social media than before and even perform daily activities on them. Because of the convenience of social media services, activities such as online shopping can be easily done on social media. Social media commerce has been one of the popular ecommerce trends in recent years. AI applications in customer services such as chatbots and personalization help business to understand their customers better and improve customer experience. Intent classification is one of the techniques used in these applications. This paper focuses on the classification of users’ intentions based on the user comments posted in Myanmar Language on social media shopping pages. Convolutional Neural Network (CNN) is applied to classify the users’ comments to one of the predefined intent categories. According to the experimental result, intent classification model with name normalization plus word segmentation preprocessing can give the F-score value of 86.6.