Triple judgment method and device based on knowledge graph background information

The invention relates to a triple evaluation method and device based on knowledge graph background information. The method comprises the following steps: inputting a target knowledge graph triple data set and a knowledge graph background matrix obtained by last iteration updating into a constructed evaluation network, and when a loss function of the evaluation network is not less than a first preset threshold value, randomly selecting an unmarked real triple from an unmarked triple data set, missing out-degree information of the out-degree background matrix is obtained according to the node relation and the tail node in the unmarked real triad, and missing in-degree information of the in-degree background matrix is obtained according to the node relation and the head node in the unmarked real triad; and according to the missing out-degree information and the out-degree background matrix, obtaining a currently iteratively updated out-degree background matrix, and according to the missing in-degree information