Knowledge Graph Technology-based Active Research and Judgment Technology for Electric Power Customer Complaint Risk

Aiming at the current problem of slow feedback of electric power fault diagnosis and repair, the research constructs a knowledge graph with fault information in the text of customer complaint information. It also integrates the knowledge graph and active research and judgment technology to construct an active research and judgment system for electric power fault customer complaint information. The system utilizes two-way encoder and conditional random field model in the knowledge graph module to construct fault models and extract textual relations. D-S evidence theory is used in the research and judgment technique for information matching. The experimental results show that the knowledge graph constructed by the study has high accuracy rates of 0.975 and 0.951 in entity extraction and relation extraction, which is higher than the extraction performance of other models. In power fault diagnosis, the diagnosis correct rate of the research proposed method is 96%, which is higher than that of other algorithmic models. The experiment proves that the grid fault complaint feedback system fused with knowledge graph and research and judgment technology constructed by the research has good extraction efficiency and diagnosis accuracy. The method proposed in the study improves the efficiency of power grid fault diagnosis and takes into account the actual life experience of power grid users. Therefore, it has excellent prospects and application benefits in major power security scenarios.