Towards a Model for Personality-Based Agents for Emotional Responses

• Rafael Guimarães Rodrigues CEFET-RJ
• Gustavo Paiva Guedes CEFET-RJ
• Eduardo S. Ogasawara CEFET-RJ

Resumo

Affective Computing is a promising research area with many open challenges. This area expects to develop computational systems that can monitor and respond to the affective states of an interacting user (IU). These affective states can be observed in terms of emotional responses, which demands continuous monitoring of IU. However, emotional responses can vary according to differences in personality traits. Due to that, we propose a novel model for personality-based agents to produce different emotional responses. This model suggests that distinct personality-based agents could have particular emotional responses. To evaluate our model, we collected data from an on-line social network in which entries have personality expressed by users and emotional information associated. We produced two personality-based agents: one extroverted and another introverted. Experimental results indicate that our model can be promising in providing distinct personality-based agents. The extroverted agent performed better on texts written by extroverted individuals just as the introverted agent performed better on texts written by introverted individuals. Our achieved results point out that personality traits are relevant to produce emotional agents.