[Purpose/significance] Red cultural resources are produced in the process of the Chinese nation and the Chinese Communists' pursuit of truth. Constructing "red memory" by organizing and mining knowledge of red cultural resources can not only enhance national self-confidence and cohesiveness, but also be an important part of cultural self-confidence. There may be many problems when using red cultural resources, such as wide distribution, multiple sources and types, limited content and low degree of organization. In order to make full use of red cultural resources, this paper constructs a "red memory" knowledge graph based on multi-source heterogeneous data. [Method/process] Firstly, this paper constructed a red cultural resource ontology library for knowledge modeling of “red memory”. Secondly, it analyzed the composition and characteristics of red cultural resources collected through multiple channels and extract entities, attributes, relationships. Finally, the "red memory" knowledge graph was constructed through knowledge fusion and storage.[Result/conclusion] By constructing the "red memory" knowledge graph, it is possible to mine deep relationship on multi-source heterogeneous red cultural resource data, improve the organization degree of red cultural resources, and realize of intelligent services of red cultural resources.