Characteristic Extraction Method of Power Grid Infrastructure Engineering Based on BERT Model
Due to the involvement of multiple factors in power grid infrastructure engineering,it is difficult to ensure the F1 value when extracting its knowledge graph features.Therefore,this article proposes a method for extracting knowledge graph features of power grid infrastructure engineering based on the BERT model.Based on the RDF model,a knowledge graph model for power grid infrastructure engineering was constructed with triplets as the basic unit;By establishing connections between relationships(i.e.edges),effective extraction of target features in the knowledge graph of power grid infrastructure engineering can be achieved.In the test results,the design method showed that for different features,the F1 value of the corresponding extraction results remained stable at 0.90 or above,with a maximum value of 0.94.Compared with the control group,it had significant advantages in stability and effectiveness.
BERT modelknowledge map of power grid infrastructure engineeringfeature extractionengineering entityassociation relationstriaddata analysis
万方数据
维普
