基于大语言模型构建面向能源数字网络的知识图谱OA

The energy industry will accumulate a considerable amount of business data internally,and automatically mining the information in the business data will promote the improvement of relevant departments'business capabilities and reduce the huge operation and maintenance costs in the industry.The professional terminology in the energy field is complex and the entity relationships are dynamically changing.Cross domain knowledge fusion needs to solve the problem of unified representation of multimodal data(text,temporal,spatial).In response to the above issues,this paper proposes a framework for constructing an energy knowledge graph based on a large language model,designs a"patiotemporal functional"dual dimensional ontology model,and introduces digital twin technology to achieve dynamic knowledge updates.Domain adaptive prompt function was developed,accurate extraction of professional terms was achieved with Qwen model.A hybrid architecture of graph neural network Transformer was proposed to solve the problem of semantic alignment in multimodal data.The research results can provide a highly reliable knowledge base for energy digital networks and assist in the construction of new type of power system.