基于知识空间表征的习题难度模型及个性化习题推荐研究

做习题能够给学习者带来刺激和反馈，是教学过程中关键的一环。随着在线教育的发展，学习者能够获取海量的习题资源，因此引发了信息过载导致学习针对性不强、效率不高等问题，使得个性化习题推荐成为当下教育研究的热点问题。作为个性化习题推荐的重要依据，习题难度对于引导和激励学习者至关重要。鉴于此，本课题拟研究基于知识空间表征的习题难度模型及个性化习题推荐，具体包括基于知识空间表征的习题难度建模方法、基于认知诊断模型的Q矩阵修正技术、基于多源数据的课程先决依赖关系知识图谱构建技术、基于习题难度模型的个性化习题推荐技术及有效性评估策略，并研究与实现一个基于习题难度模型的个性化习题推荐原型系统。以期解决现有研究工作存在的对习题难度建模分析不足、习题难度模型受主观因素影响和个性化习题推荐没能同时考虑习题的主观难度及客观难度这三个重要问题。本课题的研究将提升个性化习题推荐的有效性，具有重要的理论价值和应用前景。

Doing exercises bring excitement and feedback to learners, which is a key component in teaching process. With the development of online education, learners can easily obtain massive exercises, which leads to the low pertinence and efficiency learning problem caused by information overload due to massive exercises. Therefore, personalized exercise recommendation becomes a hot research problem in current education research domain. As an important consideration factor of personalized exercise recommendation, exercise’s difficulty is very important for guiding and motivating learners. In view of this, this project studies the exercise’s difficulty model based on exercise’s representation in knowledge space and personalized exercise recommendation. In specific, the research content of this project includes the building method of exercise’s difficulty model based on exercise’s representation in knowledge space, the Q-matrix correction technology based on cognitive diagnosis model, the construction technology of course knowledge graph containing prerequisite dependencies among knowledge points based on multi-source data, the personalized exercise recommendation technology based on exercise’s difficulty model and the strategy to evaluate the effectiveness of the proposed recommendation technology. Besides, a personalized recommendation prototype system based on exercise’s difficulty model is also studied and implemented. The research objective of the project is to tackle with three important problems existing in related current research works that have not been well solved, namely the insufficiency in modeling and analyzing exercise’s difficulty, the impact of subjective factors to the quality of exercise’s difficulty model and the failure of considering both of the subjective difficulty and objective difficulty of exercises in personalized exercise recommendation. The research works conducted in this project will enhance the effectiveness of personalized exercise recommendation, having important theoretical value and application prospect.