Abstract: Ancient classical dictionaries are important cultural heritages of collections, and semantic association and knowledge discovery of ancient Chinese words can be realized by organizing the knowledge in them. Knowledge graph is a significant knowledge organization technique in artificial intelligence. The dictionary definition of knowledge graph is composed of entities, concepts and their relationships, which can realize intelligent retrieval and knowledge discovery. This article analyzes in depth the style and content structure of ancient dictionaries, designs the lexicographical knowledge ontology based on lexical semantics, termontography and resource description framework standards, and puts forward an idea to build a model through lexicographical knowledge graph on the basis of the structuralized knowledge representation of lexicon, concept and definition. According to this model, this article conducts the extraction of vocabulary concepts, relations, definitions and examples, finishes the transformation, storage and dissemination of associated data in the meantime. Finally, taking the classical dictionary Erya as an example, the article expounds on the construction process of knowledge graph of ancient dictionaries and its application in semantic search, intelligent learning, digital humanities and other scenarios.