Liu Xindi

Zibo Polytechnic University

Abstract：

With the explosive development of Artificial Intelligence Generated Content (AIGC), represented by Large Language Models (LLMs), the ecology of knowledge production and dissemination in universities is undergoing a disruptive reconstruction. Traditional University Knowledge Management (UKM) faces challenges such as "data silos," "passive retrieval," and "low structured utility." This paper explores the paradigm shift of UKM from "Stock-Flow" to "Generative-Iterative" under the empowerment of AIGC. We propose a "Neuro-Symbolic" dual-drive UKM framework that integrates Knowledge Graphs (KG) with LLMs to enhance the accuracy and reasoning capability of knowledge services. Through a case study of a research-oriented university's "AI-Scholar" system, quantitative analysis reveals that the AIGC-driven model reduced the average literature review time by 65% and improved the knowledge innovation index by 0.42. Finally, this study outlines action paths encompassing technical architecture, data governance, and ethical boundaries, providing a theoretical reference and practical guide for the digital transformation of higher education institutions.

Key Words：

AIGC; university knowledge management; paradigm shift; neuro-symbolic AI; digital transformation