Abstract: The paper aims to enhance the efficiency and accuracy of retrieving historical knowledge of the Communist Party of China by constructing a dual-layer semantic indexing system based on a knowledge graph. This study explores methods to unearth fine-grained Party history knowledge scattered across multi-source and heterogeneous information resources, categorizing it according to knowledge characteristics. It employs deep learning techniques for knowledge extraction to build a Party history knowledge graph layer, focusing on the design and implementation process of a dual-layer semantic indexing system tailored for Party history research papers. This study leverages the knowledge relevance of knowledge graphs to provide users with highly matched document retrieval results tailored to their needs, achieving a critical breakthrough in transforming linked data into indexes. It offers valuable insights for innovatively utilizing knowledge graph technologies to drive the digital transformation of Party history research.