A Therapeutic Dialogue Agent for Polish Language

Abstract

Mental disorders affect large numbers of people worldwide. Computer-aided therapies have been designed and successfully applied in the context of mental health, including solutions based on artificial intelligence (AI). Application of AI in mental disorder therapies frequently has the form of dialogue systems. The long-term goal of this research is to develop a therapeutic chatbot capable of having a conversation in Polish. The initial work was focused on literature review and choosing the software framework for the dialogue system. As a next step, we have prepared a dataset of emotionally grounded dialogue turns in Polish, by applying machine translation to an existing resource for English. The resulting dataset was used for training an emotion recognition model which will be incorporated into the dialogue system. Next, the chatbot will be exposed to a group of test users, and its language understanding capabilities will be further developed. In future, the system will be equipped with a speech interface to also enable verbal communication. We plan to conduct experiments with the help of therapists and patients, followed by thorough analyses of their interactions with the chatbot.