Knowledge graph completion method based on context awareness

The invention provides a context-aware knowledge graph completion method, which comprises the following steps of: calculating weights of neighbor edges of different entities according to knowledge graph sub-graphs of entity pairs and an edge-aware relation message passing mechanism, and performing weighted combination on states of relations by utilizing the neighbor relation weights to obtain neighbor edge representations of the entity pairs; the context-aware path attention mechanism learns relation paths between the entity pairs to obtain representations of all relation paths between the entity pairs; calculating the weight of each relation path according to the neighbor edge representation of the entity pair, and performing weighted summation on the representation of the relation paths by using the weights to obtain the representation of the relation paths of the entity pair; and calculating the probability distribution of the relationship between the entity pairs, and constructing a loss function for trai