Sentiment Classification of Tourism Based on Rules and LDA Topic Model
Bingyang Chen, Lulu Fan, Xiaobao Fu · 2019 International Conference on Electronic Engineering and Informatics (EEI) · 2019
Mastering the sentimental state of tourists can provide decision-making reference for scenic spot management and business operation. The core of tourist sentiment analysis is the construction of tourism sentiment classification model. At present, there is less research on emotions in the field of tourism, and the classification accuracy of the existing tourism sentiment model needs to be improved. Therefore, a method of combining sentiment lexicon and machine learning to construct tourism emotion model is presented. The tourist texts were collected from micro-blog and travel website reviews to construct tourism emotional lexicon firstly. By comparing the score of the single sentence text sentiment with threshold, the sentences with the clear emotion polarity were extracted, and we constructed a classifier preliminary with NB algorithm. Then the paper used the LDA topic model to correct the sentiment classification model, and the new model was used to classify of the other sentences. Finally, the two results were integrated as the end classification result. Through experiments on the Lijiang travel text, it is found that the accuracy of the hybrid method increased 13.85% than the sentiment lexicon method, and increased 8.51% than in machine learning.