With the in-depth development of education informatization, the optimal construction of co-curricular heterogeneous course resources and the recommendation of personalized learning paths have become an important research field in the education sector. In this paper, we propose an algorithm for the optimal construction of heterogeneous course resources and learning path recommendation based on subject knowledge mapping. Firstly, based on the linear chain conditional random field model and the relational multi-classification model, the recognition of named entities and the extraction of entity relations are accomplished respectively. Then the relationships between the obtained subject knowledge points are inputted into Neo4j graphical database to complete the visualization of subject knowledge graph. The smart learning E-GPPE-C model and the constructed subject knowledge graph are combined to construct the learner portrait and complete the design of the learning path recommendation algorithm from four cognitive levels. The entity recognition and relationship extraction models in this paper both obtain more excellent experimental results, and the learning path conforms to the law of gradual progression in education, obtaining much higher than the comparative algorithm P@20 value, which reaches 75.45. The method in this paper can effectively explore more scientific learning paths, provide a method for the optimization of the resources of co-curricular heterogeneous courses, and provide support for personalized education.
