Aplikasi Mobile untuk Identifikasi Emosi Manusia Berbasis Teks pada Jejaring Sosial Twitter dengan Klasifikasi Decision Tree

Abstract

Humans are social beings that need to communicate with others. Humans in communication often reveal emotions either directly or indirectly. Awareness of people’s emotions when communicating is one of the reasons of emotion recognition. Emotion recognitions divided into 3 categories, namely physiological signals, facial expressions, and voice. However the existing method has limitations in use or cost. A potential solution is affective computing, which can recognize emotion through typing patterns by collecting and analyzing user behavior and context from touchscreen, accelerometer, and GPS on the smartphone. This method was implemented as an Android app on Twitter social networking service. This research used decision tree classification method to identify 7 types of emotions i.e. happiness, surprise, anger, disgust, sadness, fear, and neutral by using 10 features. The data used in this study consisted 300 datasets. The highest accuracy in the research produced 44%.