面向智能教育的自适应学习关键技术与应用

Key techniques and application of intelligent education oriented adaptive learning

本文是关于我们获得2020年度吴文俊人工智能科学技术奖主要工作的一个介绍。该成果针对自适应学习中面临的教学资源表示困难、学习状态诊断困难以及学习策略设计困难等关键技术难题,首先构建数据驱动的教学资源无监督表示新框架,提高了教学资源质量评估和内容检索的精度和效率。其次提出基于深度学习的学习者认知诊断新方法,突破了以量表为基础的教育测量理论研究范式。然后设计基于知识匹配的个性化推荐技术以及多目标匹配的自适应推荐技术,满足了智能教育场景的复杂约束与学习者的多样目标需求。最后,本文成果研发了面向基础教育的智能教育系统——智学网,已在全国推广使用,对我国智能教育发展具有积极意义。

This paper is an introduction to our main work in winning of the 10th Wu Wenjun Artificial Intelligence Sci-ence and Technology Award.In response to key technical problems in adaptive learning,such as the difficulties in the representation of teaching resources,the diagnosis of knowledge state and the design of learning strategies,we first con-structed a new framework for data-driven unsupervised representation of teaching resources,which significantly im-proves the accuracy and efficiency of quality evaluation and content retrieval of teaching resources.Then proposed a new method for learner cognitive diagnosis based on deep learning,which breaks through the scale-based research paradigm of educational measurement theories.And further designed a personalized recommendation technology based on knowledge matching and an adaptive recommendation technology based on multi-objective matching,meeting the complex constraints of intelligent education scenarios and the diverse learning targets of learners.Finally,an intelligent education system—“Zhixue.com”was developed for basic education,which has been popularized and used throughout the country.This system plays a positive role in the development of China’s intelligent education.