计算机应用研究2026，Vol.43Issue(6)：1835-1844,10.DOI:10.19734/j.issn.1001-3695.2025.09.0366
基于认知流畅性的课程思政素材生成方法
LLM-driven generation of ideological and political materials for curriculum
摘要
Abstract
Large language models(LLMs)demonstrate significant potential in curriculum ideological and political education,assisting teachers in rapidly generating ideological and political education cases integrated with value guidance.However,the challenge of aligning"knowledge points with ideological and political elements"limits their application.In response to this challenge,this paper proposed a"generator-evaluator"framework based on cognitive fluency and constructed an IPEC-GE dataset comprising 839 high-quality samples.The generator produced ideological and political education cases following a structured chain:"case introduction → knowledge linkage → ideological sublimation"guided by specific prompts.Two evaluators assessed the two dimensions of cognitive fluency:logical consistency and semantic coherence.Through collaborative optimization,the generator and evaluators enhanced the quality of generated content and resolved the alignment issue between"knowledge points and ideological and political elements."Experimental results show that CF-IPG method successfully gener-ates ideological and political education cases for 100 knowledge points in Data Structure,achieving a human evaluation score of 0.902 1.Through dual-dimensional evaluation of cognitive fluency and a collaborative training mechanism,it effectively achieves deep alignment between knowledge points and ideological and political education elements,providing an efficient and feasible technical way for the large-scale application of course-based ideological and political education.关键词
课程思政/认知流畅性/大语言模型/思政素材/知识点-思政要素对齐/生成器-判别器架构Key words
curriculum ideological and political education/cognitive fluency/large language model/ideological and political materials/knowledge point-alignment of ideological and political education elements/generator-evaluators framework分类
信息技术与安全科学引用本文复制引用
何绯娟,王鑫平,夏悦,姬志国,陈皓然..基于认知流畅性的课程思政素材生成方法[J].计算机应用研究,2026,43(6):1835-1844,10.基金项目
国家自然科学基金面上项目(62477037) （62477037）
陕西省社会科学基金资助项目(2024P041) （2024P041）
陕西高校青年创新团队"多模态大数据挖掘与融合创新团队"资助项目 （）
