A Goal-Oriented Chatbot for Engaging the Elderly Through Family Photo Conversations
The goal-oriented system analyzes conversations to identify favorite topics and reduce loneliness.
A research team led by Raymond Chung, Keith Ng, and CD Shum has developed a novel AI chatbot system designed to combat loneliness and stimulate cognitive function in the elderly. The core innovation is a goal-oriented dialogue framework that uses family photos uploaded by caregivers as conversation starters. The chatbot guides interactions by asking structured "W" questions—who, where, when, and what—about the photo to engage memory, followed by an open-ended question to encourage positive reminiscence.
After each conversation, the system performs a key analysis: it identifies topics the user favored or disliked during the chat. This allows it to intelligently suggest the next photo, either featuring the same family members or an individual previously mentioned, creating a personalized and evolving interaction loop. The research, accepted at the 2025 IEEE COMPSAC conference, also includes a dedicated web portal for caregivers. This portal is not just for uploading photos; it allows caregivers to review chat logs, giving them a unique window into the user's emotional state and preferences, turning social interaction into a measurable well-being metric.
- Uses a goal-oriented dialogue framework to structure conversations around family photos, starting with specific "W" questions.
- Analyzes chat content post-conversation to identify user preferences and automatically suggest the next relevant photo.
- Includes a caregiver web portal for photo uploads and conversation review, providing direct insights into user well-being.
Why It Matters
It transforms passive photo viewing into an active, therapeutic tool for cognitive engagement and loneliness reduction in aging populations.