Thai Comments Sentiment Analysis on Social Networks with Deep Learning Approach

Sentiment analysis is a kind of a Natural Language Processing (NLP) tasks in text classification used to determine the sentiment polarity from any given document. In this research, two problems were solved. The first problem is the word ambiguous manipulation. Another task is an automatic sentiment classification. This work shows our process of Thai online document cleaning to attack the uncertainty in Thai text. This work also includes attempts to find the appropriate deep learning algorithm to classify the sentiment of Thai online document by comparing the performance among several approaches. The results show that a model using the bi-directional gated recurrent unit (GRU) with attention mechanism yields the best performance in sentiment classification.