Abstract: This paper mainly focuses on the application of network security knowledge graph, combined with embedded representation technology, and proposes the highest relevant search method based on keywords. The network security knowledge graph with vulnerability as the core was constructed by using the open network security vulnerability data. On this basis, the highest relevant search method based on keywords was designed by combining embedded representation technology and Steiner tree algorithm of large graph. The improved Trans-W series embedded representation algorithm with relation weight was proposed based on the network security features. At the same time, SketchTrim algorithm suitable for large graphs is proposed based on Sketch. Experimental results show that the accuracy is significantly improved.