Exploration and Practice of Blended Teaching Reform of the Analog Electronic Technology Course Driven by an Artificial Intelligence (AI) Smart Knowledge Graph

Abstract

The Analog Electronic Technology course is concept-intensive, and its circuit analysis methods are flexible. A relatively common difficulty among students is that “single-point knowledge can still be mastered, but the overall context is difficult to connect.” In teaching, problems such as linear content organization, unified learning support, a long feedback cycle, and mismatch between resources and needs have existed for a long time. In response, this study takes Outcome-Based Education (OBE) as the basic guidance, connects the structured organization ability of the smart knowledge graph with the dialogue interaction ability of the artificial intelligence (AI) agent, and constructs a blended teaching model driven by an AI smart knowledge graph. Starting from graduation requirements and course objectives, knowledge points are reversely decomposed, four types of associations, namely prerequisite, parallel, causal, and application associations, are sorted out, and a three-layer smart knowledge graph organized in five levels of “course-module-chapter-knowledge point-sub-knowledge point” is formed. On this basis, an AI teaching agent is developed to undertake preview guidance, in-class question answering, homework correction, learning situation analysis, and other work. The teaching process is then rearranged according to the three stages of before class, in class, and after class. A one-semester comparative practice shows that the final average score of the experimental class is 8.7 points higher than that of the control class, the score difference in comprehensive questions is especially obvious, the proportion of students independently completing design experiments has increased, and the dispersion of scores has also narrowed. The relevant practices have certain reference value for the intelligent transformation of basic courses in electronic information majors, but the scope of application of the conclusions is still limited by the sample size and implementation rounds.