深度学习的目标与评价体系构建

Study on the Evaluation Theoretical Structure Building of Deep Learning

深度学习的目标面向提升学习者的问题解决、高阶思维、自主学习和知识创新等高阶能力。深度学习评价应以深度学习目标为导向，运用调查、测验、统计分析等方法，对深度学习过程及结果做出价值判断，对深度学习目标进行反思和修订。该文就如何评价深度学习这一问题，提出构建以布鲁姆的认知目标分类法、比格斯的SOLO分类法、辛普森的动作技能目标分类法和克拉斯沃尔的情感目标分类法为基础的深度学习多维评价体系，以非结构化的深层知识、高阶认知技能、高阶思维能力和高水平动作技能等的形成为深度学习评价的现实标准，构建认知、思维结构、动作技能和情感四位一体的深度学习评价体系，以解析不同领域中深度学习者可达成的预期目标。

Deep learning orients to develop learners’ high-order abilities， such as problem solving， high-order thinking， independent learning and knowledge innovation. The Evaluation of deep learning is based on the objectives of deep learning， using the methods of investigation， test and statistical analysis， to make a value judgment about the process and outcome of deep learning， to reflect and revise the goals of deep learning. In order to illustrate how to assess deep learning， this paper points out that the intended objectives of deep learning is to generate unstructured deep-knowledge， high-order cognitive skills， high-order thinking abilities and high-level motor skills. Accordingly， it proposes to build the four-in-one evaluation system of cognition， thinking structure， motor skill and emotion. That is to take the Bloom’s taxonomy of cognitive objectives， Biggs’ SOLO taxonomy， Simpson’s taxonomy of motor skills and Krathwohl’s taxonomy of emotional targets as the theoretical basis of deep learning evaluation.