New Approach of Diagnosing Sarcasm using Multi Action Units and Head Movements

Abstract

Sarcasm, a sharp and ironic utterance designed to cut or to cause pain, is often used to express strong emotions, such as contempt, mockery or bitterness. Sarcasm detection is of great importance in understanding people's true sentiments and opinions. Most Existing techniques of detecting sarcasm use speech, writing text, signs as a base to catch percent of sarcasm even with limited accuracy. In this article, a new approach of detecting sarcasm using three techniques of machine learning has been proposed. Eight action units and eight head movements from Facial Action Coding System are considered as a base rule to generate effective features to train and test machine learning techniques. The selection of these units and movements are performed according the advice of some specialist psychiatrists. The simulation results view significant accuracy of diagnosing sarcasm exceed 86% which reflects the robustness of the proposed approach also open the way for further works in this field of research.