LINE Chat Record Emotion Classification Using SVM

Abstract

There is abundance of precious information that can be tapped from the Social network sites. Much research has been done using data mining to tap information that can be used to forecast user emotion. It can also be used in user behavior forecasting like pop forecasting, stock analyses and more. However, most information is easily collected from the public domains. Most research is focused on public information analysis while very little research is focused on the private less accessible information. In this paper, the public post records from twitter are used to analyze private chat record from LINE. Training of the data is performed using support vector machine (SVM). The network public data are training data and the personal private data are testing data for emotion classification. Results showed that the public data used to analyze private data is feasible and resulted in significant level of accuracy.