Electric power professional knowledge graph automatic completion method based on graph neural network

The invention relates to the field of artificial intelligence and electric power. The invention discloses an automatic completion method for an electric power professional knowledge graph based on a graph neural network, and the method employs the graph neural network to represent an out-of-graph entity, brings the out-of-graph entity into an electric power professional artificial knowledge graph frame, and achieves the automatic completion of an artificially established electric power professional knowledge graph. According to the method, the content of the manual knowledge graph of the electric power major is enriched, retraining of entities outside the graph can be avoided, the operation burden of machine equipment is relieved, the learning cost is reduced, the existing knowledge features are learned through the TransE algorithm, the hidden relation between the knowledge entities is mined, and the function that the manual knowledge graph of the electric power major executes automatic complementation accord