MULTI-MODAL FUSION UPDATING METHOD FOR KNOWLEDGE NODES

Provided in the present invention is a multi-modal fusion updating method for knowledge nodes. The method comprises: determining knowledge modality types included in knowledge data; according to a knowledge modality type of each knowledge node and the amount of knowledge information of the corresponding knowledge modality type, performing calculation to obtain a modality evaluation sub-coefficient, and obtaining the average modality evaluation coefficient according to modality evaluation sub-coefficients of all knowledge nodes; sorting first knowledge nodes in an ascending order according to the modality evaluation sub-coefficients, so as to obtain a knowledge node sequence; selecting from the knowledge node sequence a plurality of first knowledge nodes at the beginning as second knowledge nodes, and according to knowledge modality types of the second knowledge nodes at the current moment, generating recommended modality types to be added that correspond to the second knowledge nodes; and according to a confi