Lexicon-based emotion classification on movie reviews.

Abstract

In this study, an emotion tagging program was developed for movie reviews. The emotion tagging program can be used to enable a movie buff or any user to browse for movies using emotion words. To develop an emotion tagging program, emotion lexicons for emotion classification are needed. Emotion classification focuses on the strength and intensity of emotions. An efficient way to recognize emotion expressions is to identify obvious emotion words in texts. The emotion lexicons in this study were collected from three public online resources: WordNet-Affect, General Inquirer and Roget's Thesaurus. These emotion words were classified according to two perspectives: emotional and semantic. From the emotion perspective, the words were classified into Ekman's six basic emotion categories. These emotion categories were further divided into seven other categories according to the semantic concepts of the words, using the ideas of two researchers, Johnson-Laird and Oatley.