Using subjective emotion, facial expression, and gaze direction to evaluate user affective experience and predict preference when playing single-player games
提出一种基于愉悦-唤醒-支配情绪、面部表情和注视方向的方法，评估单机游戏用户的情感体验，并构建人工智能模型预测用户偏好，实验表明该方法在不同难度下有效。
The affective experience generated when users play computer games can influence their attitude and preference towards the game. Existing evaluation means mainly depend on subjective scales and physiological signals. However, some limitations should not be ignored (e.g. subjective scales are not objective, and physiological signals are complicated). In this paper, we 1) propose a novel method to assess user affective experience when playing single-player games based on pleasure-arousal-dominance (PAD) emotions, facial expressions, and gaze directions, and 2) build an artificial intelligence model to identify user preference. Fifty-four subjects participated in a basketball experiment with three difficulty levels. Their expressions, gaze directions, and subjective PAD emotions were collected and analysed. Experimental results showed that the expression intensities of angry, sad, and neutral, yaw angle degrees of gaze direction, and PAD emotions varied significantly under different difficulties. Besides, the proposed model achieved better performance than other machine-learning algorithms on the collected dataset.