Analysis of convolution neural network for transfer learning of sentiment analysis in Indonesian tweets

Sentiment analysis is an activity to classify public opinion about entities in textual data into positive or negative. One of the automatic methods for sentiment analysis is convolution neural network (CNN). CNN consists of many layers with many parameters that can be adjusted as needed to form a specific architecture. CNN works well in similar domains; however, it gives less accurate in different domains. Therefore, we consider transfer learning which transfers knowledge from a source domain to different but related target domains. In this paper, we examine parameter sensitivity and accuracy of CNN for transfer learning of sentiment analysis in Indonesian tweets. Our simulation shows that the parameters are very sensitive and incremental learning significantly increases the accuracy of transfer learning of the CNN model.