Railways are a key infrastructure that the country is vigorously developing and a popular mode of transportation, holding an irreplaceable position in the national integrated transportation system. Over the past decade, the railway industry has actively promoted the construction of informatization, closely following the times and completing the transition to intelligence and informatization. In the process, China Railways has accumulated vast amounts of structured, semi-structured, and unstructured data. The data accumulated in the railway industry is characterized by large volumes, diverse data types, rapid growth, and high business value. The enormous volume of data and its rapid growth have resulted in decreased efficiency in the use of information and data in the railway industry, with the problem of data being present but difficult to retrieve becoming increasingly prominent. There is a lack of effective organizational methods to support comprehensive and efficient retrieval of data information within the industry. Knowledge graphs, leveraging their distinctive features, have shown extremely high application value in reducing enterprise data management costs, enhancing the efficiency and sharing of information resources, and reducing the costs of transforming data into knowledge. Additionally, knowledge graphs have demonstrated strong performance in data organization and recommendation systems. This paper proposes a feasible method for constructing a railway data resource catalog based on knowledge graphs, combining technologies such as semantic association, natural language processing, and graph convolutional networks, to support the semantic integration and intelligent retrieval of railway data.