Research and Application of Power Market Policy Question-Answering System based on Local Knowledge Base

With the growth of global energy demand and the increasing complexity of power markets, quickly acquiring and understanding policy information has become a key challenge. This paper proposes a method to use triple knowledge graphs to construct a dataset for fine-tuning a large language model and build a local knowledge base to form an intelligent question-answering system for power market policies. The knowledge graph is constructed by extracting key relationships in the power market policy text, and efficient question-answering is achieved by combining an optimized large language model. Experimental results show that the system exhibits high recall and precision in the evaluation of multiple embedding models, providing an efficient solution for the intelligent interpretation of power market policies.