AGENDA

Oct
12
11:40 - 11:50
Empowering Sustainability through AI: A Four-Dimensional Closed-Loop Practice in University Museums
Zhao Ke¹,² , Chen Deli ¹& Lv Hongyin²
¹Electronic Science and Technology Museum, China, ²University of Electronic Science and Technology,China

zhaoke@uestc.edu.cn

University museums advance the Sustainable Development Goals (SDGs), but growing collections and increasing audience demands overload their traditional manual workflows and isolated data systems. While artificial intelligence (AI) offers tremendous potential to overcome these limitations, a key challenge is how to transition AI from a novel concept into a stable, sustainable tool for university museums. Here, we present a four-dimensional AI closed-loop system used at the Electronic Science and Technology Museum (ESTM), which aims to support sustainable museum management and education. The system integrates four modules: a YOLOv8-based automated collection classification, a Large Language Model (LLM)-driven Knowledge Graph, an on-premises Hybrid RAG AI teaching platform, and a CNN-based audience micro-expression evaluation system (SMEM). The collection classification system and the Knowledge Graph work together to address the issues of heavy manual cataloguing and isolated knowledge. The local AI teaching platform ensures data security while supporting personalized learning. Additionally, capturing visitors' micro-expressions provides objective data to evaluate exhibition effects, which can effectively replace traditional questionnaires. Practical experience shows that linking these four AI modules creates an automated feedback loop. This continuous process spans from collection preservation to audience engagement, helping to increase the museum's working efficiency and educational output. This closed-loop model can also offer a practical reference for other academic museums. By applying AI to everyday work, university museums can better handle resource limits, improve education, and support global sustainable development.