Sentiment Analysis on News Headlines of Nation’s Capital Relocation Using CNN and SVM

The relocation of the nation's capital is an important decision. In Indonesia, this is seen as an urgency that aims to elevate economic growth, improve economic equality outside capital city (Java Island). The plan of moving the country’s capital has announced by various online news media and resulted many pros and cons. This research is a sentiment analysis of news headlines on the topic of moving new country’s capital of Indonesia on online news detik.com using CNN and SVM models. This research aims to know the results of sentiment and which model that have the best performance. The research flow uses SEMMA method. The 3.5K news headline dataset had 48% neutral sentiment, 34.7% positive sentiment, and 17.3% negative sentiment, which were automatically labeled by Vader. The labeled dataset from each class was then back-translated and adjusted to achieve a balanced amount of data. The dataset is fed into CNN and SVM models with Stratified K-Fold Cross Validation (k=10). Model evalution shows SVM model has better performance with 90.65% accuracy, compared to CNN with 84,92%. The sentiment result is interpreted in general and generate suggestion that the Indonesian government able to focus on the development of IKN infrastructure first to provide important signals for investors as seeing the government's seriousness in working on huge projects, but they still have to look after other priority programs so that overall government plans able to run on target.