Intelligent Evaluation and Decision Support for Curriculum Ideological-Political Education via Association Rule Mining and LLM-Enhanced Dynamic Student Profiling
Year: 2026
Type: conference-paper
Abstract: Curriculum Ideological-Political Education (CIPE) aims to integrate values education into disciplinary courses, yet current evaluation remains predominantly subjective and lacks data-driven decision support. This paper proposes an intelligent decision support system that integrates Frequent Pattern Growth (FP-Growth) association rule mining with Large Language Model (LLM) semantic analysis to enable objective, dynamic CIPE evaluation. The system adopts a four-layer architecture comprising data collection, analytical modeling, decision support, and visualization interaction layers. The FP-Growth module discovers latent associations between disciplinary knowledge points and ideological-political elements from student learning transactions, while the LLM module extracts values orientation and ideological comprehension depth indicators from open-ended reflective texts via few-shot Chain-of-Thought prompting. Together, they generate time-varying dynamic student profiles that support a real-time teacher dashboard and personalized intervention recommendations. Large-scale validation on 166 chapter-level evaluation records from 51 students revealed five core findings, including a “social desirability effect” in attitude self-assessment and a four-quadrant student clustering with 64.7% falling into low-engagement categories. Inter-rater validation confirmed the LLM's annotation reliability with Cohen's$\kappa=0.747$(substantial agreement) against expert judgment. The system demonstrates how association rule mining and LLM-based semantic analysis can be integrated as a generalizable intelligent decision support framework for values-oriented educational assessment.
Institution: Guangdong Polytechnic of Science and Technology
Language: English
Subfield: Education
Field: Social Sciences
Domain: Social Sciences
SDG: Quality education
Open Access status: hybrid
OpenAlex ID: W7203920764