新工科背景下农林院校智慧教学的研究与实践

Research and Practice of Smart Teaching in Agricultural and Forestry Universities under the Background of Emerging Engineering Education

摘要: 面对“新工科”建设对传统农林类高校提出的学科交叉融合、人才培养模式改革等一系列挑战，本文提出“数据驱动、AI赋能、虚实结合”农林类高校智慧教学新生态，并以此为指导思想，设计出其具体实现方式：首先利用知识图谱(Knowledge Graph)技术挖掘及建立农业科学领域知识图谱中与计算机领域异构知识点之间的联系，重塑多维交叉融合课程内在逻辑关系；其二，依托计算机视觉(Computer Vision)和VR技术搭建高仿真虚拟仿真实训室，在农闲时间开展作物表型性状观测、病虫草害自动诊断等全时段虚实结合实训；其三，运用深度学习(Deep Learning)算法(如LSTM模型)，创建学情发展规律时空序列预测模型，面向学习过程开展多模态数据挖掘分析，完成教学资源的智能推荐以及学业预警的精准干预。

Abstract: Against a series of challenges posed by the construction of Emerging Engineering Education to traditional agriculture and forestry universities, including interdisciplinary integration and the reform of talent cultivation models, this paper proposes a smart teaching new ecosystem for agriculture and forestry universities featuring data-driven operation, AI empowerment, and virtual-real integration. Guided by this ideology, its specific implementation approaches are designed as follows. First, Knowledge Graph technology is adopted to mine correlations between heterogeneous knowledge points in agricultural science and computer disciplines and construct relevant links, thereby reshaping the internal logical relationships of multi-dimensional interdisciplinary integrated courses. Second, highly simulated virtual training laboratories are built based on Computer Vision and VR technologies to carry out full-time virtual-real integrated training activities such as crop phenotypic trait observation and automatic diagnosis of diseases, pests and weeds during agricultural off-seasons. Third, Deep Learning algorithms (e.g., the LSTM model) are employed to establish a spatio-temporal sequence prediction model for the development rules of students’ learning conditions. Multi-modal data mining and analysis are conducted throughout the whole learning process to realize intelligent recommendation of teaching resources and precise intervention via academic early warning.