Domain Knowledge graph enhancement method

With the proposal and development of the RAG framework, knowledge graphs, due to their structured data and semantically rich characteristics, play an important role as external knowledge bases within the framework. At the same time, as general-purpose question-answering systems have become relatively mature, many enterprises are now focusing on developing question-answering applications for specific domains, leading to an increasing demand for domain-specific knowledge graphs. Although there is a wealth of unstructured data in various domains and open-source general knowledge graphs, there is a lack of knowledge graphs tailored to specific domains. In response to the scarcity of domain-specific knowledge graphs, this paper conducts research on constructing such graphs. It proposes building domain knowledge graphs by extracting domain-specific data from large general knowledge graphs and enhancing the completeness and professionalism of domain knowledge graph data by constructing domain knowledge graphs from unstructured data in the field using large models. This approach aims to provide more accurate and comprehensive information support in domain-specific question-answering applications, thereby improving user experience and the efficiency of problem-solving.