Construction of Adaptive Evaluation Model of Educational Knowledge Structure Integrating Knowledge Graph and Convolutional Neural Network

In order to accurately evaluate students' learning outcomes and knowledge mastery in massive educational data, this study combines the structuring ability of knowledge graphs and the feature extraction ability of convolutional neural networks to build an evaluation model that can adapt to different students' learning paths and knowledge structures. The article first constructs a knowledge graph based on educational knowledge, entityizes knowledge points, concepts, relationships, etc., and defines the relationship between them to form a structured educational knowledge framework; then, a convolutional neural network is used to embed the entity relationships in the knowledge graph, and the local features between entity relationships are captured through convolution operations to improve the accuracy of reasoning; then, based on learner portraits, different learning paths are planned through knowledge graphs, and courses and exercises with personalized difficulty and rhythm are provided for learners to improve learning efficiency and effectiveness; the convolutional neural network is trained using historical educational data, and the model parameters are optimized through the back-propagation algorithm, so that the model can accurately predict students' knowledge mastery and learning outcomes; finally, the model is applied to actual teaching scenarios. The results show that the average student score under the adaptive model is 84, but only 71 under the traditional evaluation model. In addition, the student classroom participation rate and interaction times under the adaptive model are better than those under the traditional evaluation model. This study provides valuable experience and guidance for the construction and optimization of future educational evaluation models, and also provides new perspectives and technical means for educational practice.