基于学习诊断的个性化学习资源推荐研究

发布时间：2018-04-15 08:14

本文选题：学习诊断 + 知识结构图　；参考：《河南师范大学》2017年硕士论文

【摘要】：随着互联网技术在教育领域应用的深入,个性化学习已经成为人们关注的热点。不断快速增长的学习资源(包括线上线下资源),让学习者面临着“信息过载”和“信息迷航”的困惑。面对海量的学习资源,如何在有限的时间内,挖掘和推荐出适当的学习资源,是教育信息化领域面临的一种挑战。与电子商务领域相比,教育领域信息的挖掘与推荐有自己的特性,即学习目标特性。用户在考虑感兴趣资源的同时,还需要考虑学习目标。兴趣偏好不是资源推荐的唯一的指标,用户的学习目标也是推荐考虑的重要因素。论文基于学习诊断,研究个性化学习资源推荐方法。首先,构建知识点关系图。基于本体构建知识点之间的关系图,建立试题知识点矩阵和试题得分矩阵,根据一定的教学策略判断学生的知识点水平。针对未掌握知识点,根据拓扑排序方法获取学生个性化知识点序列。然后,针对行为数据,进行多维度分类,分析并建立其与学习效果之间的关系,从而发现学生的学习习惯和学习兴趣。第三,根据学习目标和学习偏好,建立一种学习资源推荐机制,设计协同过滤推荐算法。为避免知识点空白,当学生掌握个性化知识点序列后,指导学生进行平行知识点的学习。第四,基于学习平台产生的数据进行实验。实验表明,学习资源推荐算法能够提高推荐的准确率和效果。
[Abstract]:With the application of Internet technology in the field of education, personalized learning has become the focus of attention.The rapid growth of learning resources, including online and offline resources, makes learners face the confusion of "information overload" and "information confusion".In the face of massive learning resources, how to mine and recommend appropriate learning resources in a limited time is a challenge in the field of educational information.Compared with the field of e-commerce, the mining and recommendation of information in the field of education has its own characteristics, that is, the characteristics of learning objectives.Users need to consider learning goals as well as interested resources.Interest preference is not the only indicator of resource recommendation, and the user's learning goal is also an important factor of recommendation.Based on learning diagnosis, this paper studies the recommendation method of personalized learning resources.First of all, construct the knowledge point diagram.Based on ontology, the relationship diagram between knowledge points is constructed, the knowledge point matrix and score matrix are established, and the level of knowledge points of students is judged according to certain teaching strategies.According to the topological sorting method, the individualized knowledge point sequence of students is obtained.Then, according to the behavior data, the multi-dimension classification is carried out, and the relationship between it and the learning effect is analyzed and established, so as to find out the students' learning habits and interest.Thirdly, according to learning goals and learning preferences, a learning resource recommendation mechanism is established and collaborative filtering recommendation algorithm is designed.In order to avoid the blank of knowledge points, when students master the sequence of individualized knowledge points, the students are instructed to learn parallel knowledge points.Fourth, the experiment is based on the data generated by the learning platform.Experiments show that the learning resources recommendation algorithm can improve the accuracy and effect of recommendation.
【学位授予单位】：河南师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3

【参考文献】