||Traditional political science research usually adopts random sampling to choose a number of subjects and inquires their opinions towards certain issues. The result is then generalized to estimate the opinions of the general population. In this study, we instead adopt text mining and sentiment analysis techniques, combined with the recognition of political science knowledge, to identify Taiwanese affections toward neighboring countries.|
This study is to investigate Taiwanese affections toward China, Japan, the United States about some important issues, as these three countries are consider the most important neighboring countries. Specifically, we base on the editorials and opinion forum from newspapers in the last four years to derive major topics by using Latent Dirichlett Allocation techniques. We then use the resultant LDA model to analyze the topic and employ sentiment analysis technique to analyze the sentiment for each post in major PTT forums that discuss the relationships between Taiwan and the three countries. The sentiment of all posts are aggregated to form the overall emotion toward China, Japan, and US.