LAUSR

With the rapid growth of web content from social media, such studies as online opinion mining or sentiment analysis of text have started receiving attention from government, industry, and academic sectors. In recent years, sentiment analysis has not only emerged under knowledge fusion in the big data era, but has also become a popular research topic in the area of artificial intelligence and machine learning. This study used the Militarylife PTT board of Taiwan’s largest online forum as the source of its experimental data. The purpose of this study was to construct a sentiment analysis framework and processes for social media in order to propose a self-developed military sentiment dictionary for improving sentiment classification and analyze the performance of different deep learning models with various parameter calibration combinations. The experimental results show that the accuracy and F1-measure of the model that combines existing sentiment dictionaries and the self-developed military sentiment dictionary are better than the results from using existing sentiment dictionaries only. Furthermore, the prediction model trained using the activation function, Tanh, and when the number of Bi-LSTM network layers is two, the accuracy and F1-measure have an even better performance for sentiment classification.