Abstract: This paper used technologies such as big data, artificial intelligence, and ultra large scale pre training models to study the intelligent analysis and correlation methods of electronic document data in complex scenarios of the railway industry. The paper introduced the overall architecture and key technologies for improving the intelligence of electronic documents, which provided methodological exploration for the long-term scenario of intelligent analysis of railway electronic document data, thereby could fully tap into the value of railway electronic document resources, and improve the quality and efficiency of document processing.