<p>At present, there is a lack of fire risk assessment models and effective fire prevention measures for wooden components in historical buildings that are susceptible to fires. This paper combines the physical principles of fire spread and the basic principles of directed graphs to propose the directed graphical combined breadth-first search (DG-BFS) model, which can infer the real-time fire spread situation accurately and quickly. With the assistance of the DG-BFS model, historical building fire risk assessments are conducted using generated spread matrices, taking into account static metrics related to building parameters and dynamic factors associated with environmental conditions. Additionally, it employs node importance metrics, such as in-degree and out-degree, to evaluate the fire risk level of building nodes. In order to prevent and control the spread of the fire under the existing building, the directed graphical model employs a node deletion measure to assess the feasibility of reducing fire risk through the insulation of individual buildings. Through simulations of fire spread in actual villages in Guizhou province, the results demonstrate that the application of the aforementioned methods, combined with fire safety reinforcement of a small number of high-risk buildings, can significantly reduce the number of buildings ignited after a fire. These findings provide a method for improving fire risk assessment in historical wooden building clusters, particularly in cases with limited data, offering valuable guidance for research and practice in related fields.</p>