Construction and Application of Academic Early Warning Knowledge Graph

Aiming at the problems of insufficient measures, low process and visualization in the academic early warning system about “prevention in advance”, the knowledge graph of academic early warning was constructed and applied. Firstly, the schema layer was constructed through protégé, and the data structure of knowledge（including entities, relationships and attributes） was designed. In this step, we adopted the tree structure, so that each subclass inherits the attributes of its ancestor node. Secondly, take fact triplet as a unit to store specific information. Then, the relational database was used to realize the mapping between data and ontology, and the knowledge extraction was carried out from the structured data of the relational database. The structured data was converted into triple data through the D2RQ tool and stored in SQL. Finally, the Neo4j map database was used for visual display to complete the construction of the academic early warning knowledge graph. The experimental test results of open data sets show that the constructed academic early warning knowledge graph can give early warning to students’ studies. After sampling annotation to entities and attributes, the accuracy is 94.23%, and the timeliness is good. The system starts transmission after an average of 9 ms, and completes it after 25 ms. At the same time, the process and visualization are greatly improved, and the “prevention in advance” is realized.