The Analysis and Summarizing System of Thai Hotel Reviews Using Opinion Mining Technique

This paper proposes a development of a new opinion mining framework for analyzing and summarizing the hotel reviews on Thai language. This framework uses a hybrid concept between a keyword knowledge and a classification technique to identify a opinion feature of hotel reviews into 3 aspects (location, service and worthiness) and to classify the opinion polarity of hotel reviews into 2 classes (positive and negative). The research methodology consists of 4 steps as 1) preprocessing 2) hotel feature identification 3) opinion polarity classification and 4) hotel review summarization. Experiments showed that the accuracy of the opinion feature identification is 83.33% and the accuracy of the opinion polarity classification is 81.47% with a testing data which be manually collected from a hotel review on the agoda website.