Abstract
With the deepening of vocational education reform and its growing public visibility, media-driven public opinion has become a critical factor shaping educational governance and policy implementation. In the digital media environment, traditional qualitative approaches are insufficient to capture the scale, dynamics, and emotional complexity of news discourse. This study integrates Natural Language Processing (NLP) techniques to systematically analyze vocational education–related news public opinion and to explore sustainable governance pathways based on empirical evidence. Using authoritative news data collected over a defined period, sentiment analysis, topic modeling, and temporal evolution analysis are applied to identify emotional patterns, dominant themes, and time-specific trends. The results reveal clear structural characteristics of public opinion, with discourse concentrating on policy reform, employment outcomes, and industry–education integration, alongside sentiment fluctuations associated with major policy events. Based on these findings, the study proposes data-driven and collaborative governance strategies to enhance public opinion management and support the sustainable development of vocational education systems.