In a learning environment, emotional factors influence student motivation. Students emotion
have an important role in students' capability to learn. The tendency of the students emotion
are not easily recognizable in a short time. Twitter is a popular micro-blogging system
especially for students. Students post tweet about activities, experiences, their feelings
anywhere, anytime and in real time. Sentiment analysis on twitter produce content sentiment
that represents the feelings and emotions of students. Sentiment analysis system was built using
backpropagation method at the stage of classification. In this research backpropagation
network and the classification results were tested using WEKA with multilayer perceptron
classifier. The results of sentiment analysis with 30 student respondents are 33.33% tendency
of positive emotions, neutral emotions tendency 53.33% and 13:33% negative emotional
tendencies. The results are used as reference in providing the appropriate treatment of the
students during the process of learning.