Metinlerde duygu analizi

Abstract

Emotion Analysis in Texts aims to predict the feeling that the text gives in text-based data sets. The emotion given by the text can be positive or negative polarization shaped, or can be any emotion such as anger, joy, sadness. In this notice, both polarization analysis and four basic emotion (anger, sadness, joy and fear) analysis are examined. As a method K-Nearest Neighbor (KNN), Case Based Reasoning (CBR) and Naive Bayes methods are applied to Zolkepli's English Text Emotion (ETE) dataset. The ETE dataset is a new dataset and there is no published result on that dataset yet. For four basic emotion analysis, the highest success rate on ETE dataset is 96.4% which is obtained by KNN algorithm. For the polarization analysis the highest success rate on the ETE dataset is 99.5% and obtained by KNN algorithm after first applying CBR method to the training set of ETE dataset. © 2019 IEEE.