摘要
针对传统表现性评价在测量高阶素养时面临的“测不准”“看不全”与“用不好”三大实践困境,研究基于证据中心设计、多模态学习分析与人机协同智能理论,提出了一个整合AI能力、评价场景、评价任务与评价数据的CSTD整合模型。该模型以AI的全景式感知、循证式解析与诊断性综合三项关键能力为核心驱动,通过在智能化场景中实施结构化的表现性任务,系统采集学生在真实情境中的多模态、全过程数据。研究以劳动素养评价为例,设计并实施了项目式任务进行模型验证。结果表明,CSTD模型能有效指导AI赋能的评价实践,实现对劳动素养的自动化分析,并生成集多维度可视化、循证诊断、个性化建议与纵向追踪于一体的反馈。
To address the three persistent practical challenges in traditional performance-based assessment of higher-order competencies—namely,inaccurate measurement,incomplete observation,and ineffective applica-tion—this study aims to construct a novel AI-supported assessment paradigm.Grounded in Evidence-Centered De-sign(ECD),Multimodal Learning Analytics(MMLA),and Human-in-the-Loop AI theory,we propose an inte-grated CSTD model that synergizes AI Capability,assessment Scenario,Task,and Data.Driven by three pivotal AI capabilities—panoramic perception,evidence-based analysis,and diagnostic synthesis—the model systematically collects multimodal,whole-process data from students engaged in structured performance-based tasks within intel-ligent scenarios.Using the assessment of labor literacy as a case study,we designed and implemented a project-based task to validate the model.The results demonstrate that the CSTD model effectively guides AI-empowered as-sessment practices,enabling the automated analysis of labor literacy and the generation of feedback that integrates multi-dimensional visualization,evidence-based diagnostics,personalized suggestions,and longitudinal tracking.
作者
郑勤华
吴瑶
Zheng Qinhua;Wu Yao
出处
《教育学术月刊》
北大核心
2026年第1期13-22,共10页
Education Research Monthly
基金
国家自然科学基金面上项目“基于多模态数据融合计算的中小学生坚毅力测评技术与溯源研究”(项目代码:62277004)。
关键词
人工智能
高阶素养
表现性评价
CSTD模型
artificial intelligence
higher-order competencies
performance-based assessment
CSTD Model
