Classification of emotions in indonesian texts using K-NN method

Classification of emotions in indonesian texts using K-NN method

Abstract

This paper aims to classify texts in Indonesian language into emotion expression classes. The data were taken from 6 basic emotion classes whose training documents and test documents were obtained from articles in www.kompas.com, www.suaramerdeka.com, and www.detik.com. The text weighing was processed by using TFID method which is an integration of Term Frequency (TF) and Inverse Document Frequency (IDF). In the classification process, K-Nearest Neighbor (K-NN) was used to see how far this method could classify emotion expression of Indonesian language. The test shows that the classification of the Indonesian texts for the six basic emotion classes by using K-NN method results in accurateness percentage of 71.26%, obtained at k=40 as the optimum value. 

Keywords: basic emotions, K-Nearest Neighbor, Indonesian language, TFIDF


Human’s facial character animation is important especially in shaping face expression. This character animation is a difficult object to be animated. Facial movement in a specific pattern is called facial expression, and this is a complex matter in creating animation [1]. The emergence of intelligent agent technology has made us realize that there is an opportunity to develop an interface to improve a model of human and computer interaction, virtual character simulation for different applications such as entertainment, education, and so on. Nowadays, human and computer interaction has been done through texts, mouse, or keyboard simultaneously, and along with the rapid development in graphic computing and speech recognition technology, this interaction becomes more adaptive, flexible, and human-oriented [2]. A successful computer human interaction system should be able to recognize, interpret and process human emotions. Affective computing could offer benefits in an almost limitless range of applications. human emotion recognition is multimodal in nature, and includes textual, visual and acoustic features. Text seems to be the most studied modality since the text is relatively easier to process than others. Human Emotion Recognition from text can be simply envisioned to be a classification An agent can be said to be intelligent if it is equipped with emotion [3], therefore the agent needs to be given emotions. Meanwhile, communication can be done through verbal and non-verbal information. Verbal information can be in the form of writings obtained from words, sentences, paragraphs, etc., while non-verbal one in the form of body gesture [4].


Classification of emotions in indonesian texts using K-NN method