| （英） |
Japan currently has the highest proportion of older adults in the world, and this percentage is expected to continue increasing. Alongside this demographic shift, the number of older adults living alone is also on the rise, contributing to concerns about social isolation and loneliness.
To address this issue, we are developing a virtual agent (VA) listening service that engages in daily conversations with home-dwelling older adults. However, prior VA-based systems have struggled to generate personalized responses tailored to each user’s background and needs.
In this study, we propose and implement a method for automatically constructing user profiles based on dialogue history between the VA and the user. By leveraging Facebook AI Similarity Search (FAISS) and large language models, the system enables natural and empathetic dialogue tailored to the user's personality, hobbies, interests, concerns, lifestyle, and recent experiences. This approach improves the individualization of interactions, contributing to more engaging and supportive communication for older adults. |