Overview
Older people often struggle with mental health problems caused by loneliness, which is exacerbated by the COVID-19 pandemic. However, due to a significant imbalance between demand and supply of mental health counselling services, they cannot always get immediate help from mental health professionals. In an ongoing project, we propose to use recent digital advances to develop an emotional support chatbot system that is empathetic and can hold conversations with older people to understand and support their emotional needs. The potential usefulness of such a system to help older people cope with loneliness has been confirmed by organizations that care for older people. Our chatbot system is characterized by the following features. First, it is able to recognize emotions in conversations thanks to powerful natural language processing (NLP) techniques. Secondly, it combines the advantages of retrieval-based and generative-based chatbots in one system. Depending on the area of conversation, the chatbot can automatically respond with predefined utterances according to the context of the conversation and the emotions of older people, or intelligently generate sentences and appropriate emotions using the trained NLP models. Thirdly, the chatbot system is based on the open-source frameworks ParlAI and Rasa, which ensure language functionality and a user-friendly interface tailored to the preferences of older people. The language of the chatbot is German, which suits our target group in central Switzerland. In this talk, we will present our system architecture and the chatbot’s performance so far, as well as potential ethical issues with such an emotional support system.
