Knowledge Graph Construction Method for Power Grid Infrastructure

With the increasing of power consumption, the scale of power grid infrastructure construction is developing rapidly. A large amount of information will be generated in the power grid infrastructure construction, and traditional information management methods cannot make efficient use of this information. Knowledge graph is introduced to solve the mentioned issue. This paper focus on the concrete management in power grid infrastructure construction, introduces a power grid infrastructure knowledge graph constructing method. This method includes the following steps. Firstly, the original data of documents, tables, and news is obtained through a data collection module. Secondly, structured/unstructured data are preprocessed. Thirdly, the knowledge extraction is performed to obtain entities and relations. For unstructured data, BiLSTM-CRF is used for named entity recognition and then the BERT model is used for entity relationship extraction. Then, the technology of knowledge fusion is used to obtain the triple as entity-relation-entity. At last, knowledge is represented by storing the graph in Neo4j.