Quantitative Expression of Elderly Multi-Modal Emotions with Spoken Dialogue Agent and Edge AI

Abstract

In a super-aging society, the problem of aging of the physical and cognitive functions of older adults receiving home care is becoming more serious. In the previous study, we developed a spoken dialogue agent system that allowed older adults to communicate with machines by voice. However, it was difficult to visualize the thoughts extracted from the spoken dialog of the elderly in detail. Therefore, this study was carried out to extract the emotional characteristics of the elderly and propose a method to visualize them quantitatively. The proposed method uses AI technology to extract and express emotional features from multiple perspectives while protecting the privacy of the elderly. In this way, the method objectively reflects the personality and preferences of the elderly and leads to the prevention of dementia.