This paper first introduces ontology construction, knowledge extraction, and graph database in the process of knowledge graph construction based on folk dance. Secondly, the semantic similarity algorithm is used as the basis to construct the conceptual similarity (MD4) model, and the five types of similarity algorithms in the MD4 model are introduced. After that, based on this model, we will conduct a comparative study on the number of folk dance publications, keywords, and hotspots in China and abroad. Finally, suggestions are made for creating a knowledge graph that is based on folk dance. The annual volume of folk dance research published in China is generally increasing. In contrast, the amount of foreign publications on folk dance is generally low, and its maximum number of publications is only 67. In China, 1021 keywords finally got 156 nodes, and the total link strength of the total number of occurrences was 3942. 1352 keywords finally got 137 nodes, and the total link strength of the total number of occurrences was 1983. Among all the keywords in China, the top three co-occurrence-ranked keywords and hot words were folk dance, value, and development status, with a number of 817, 633, and 607, respectively. The foreign countries include folk dance (453), value (233), and performance (146). The cultural inheritance of folk dance can be greatly impacted by the knowledge map’s ability to summarize the development of folk dance.