This paper determines the knowledge graph architecture, collects music education teaching data from music education websites using crawler scripts, extracts knowledge from it based on the BILSTM-CRF combinatorial model, and constructs a knowledge graph of music teaching resources by storing teaching resources in the form of a graph database through knowledge fusion and knowledge processing. With the goal of improving students’ music literacy, a music teaching model incorporating knowledge mapping was designed, and finally its effect was evaluated. After 16 weeks of teaching intervention, the indicators of the control group were not significant (P>0.05), while the values of the indicators of the experimental group were significantly different (P<0.05), which proved that the introduction of the knowledge mapping on the original music teaching model was particularly effective in improving students’ music literacy.