An emotion classification model of chatbot design for the elderly

Ageing is a natural process associated with cognitive decline, an increased risk of chronic health issues, and sensory losses that have emotional consequences, leading to depressive symptoms and an increased suicide risk among the elderly. To support elderly, productive ageing activities using Chatbots offer numerous benefits. However, existing chatbots often fail to engage with elderly emotions effectively. Evoking emotional responses in users has also received less attention in previous chatbot design studies, which have mostly concentrated on usability. Conclusively, past literature has led to three problem statements: the identification of chatbot UI design elements is insufficient to influence the elderly's emotional response, there is a deficiency in addressing the elderly’s emotional needs when designing chatbots, and the influence of the elderly’s emotional responses on chatbot user interface design (UID) and their connection to the UID component have been disregarded. Therefore, this research addresses four objectives: i) to identify the User Interface design elements of the chatbot design that influence the elderly’s emotional responses; ii) to determine the emotional responses that influence the elderly’s use of the chatbot; iii) to develop an emotion classification model for chatbot design for the elderly; and iv) to validate the proposed model. This research employed a qualitative methodology using a semi-structured interview approach with ten chatbot design specimens and 30 elderly participants. Three experts were interviewed to validate the model, confirming its potential to promote positive emotional responses in elderly chatbot users. Although all of these outcomes were based on evaluations performed with some limitations and constraints, they provide some novel foundations by proposing elderly emotion classification in Chatbot UID. The key finding is the Emotion Classification Model of Chatbot Design for Elderly development, a novel approach to crafting a chatbot UI design that primarily targets elderly emotional responses. As a conclusion, the Emotion Classification Model of Chatbot Design for Elderly, makes modest theoretical, empirical, and methodological contribution that can be use as reference for UI designer to design a Chabot design based on elderly emotion needs. Future research could investigate the model's applicability using a broader range of subjects, alternative emotion dimensions, additional design elements, new domains, or integration with artificial intelligence (AI) and analytics to strengthen the internal validity of the current findings.