慕课学习环境下基于学习对象网络的个性化学习路径推荐

How to effectively help learners improve learning management, plan learning paths and implement self-directed learning is one of the key factors for succeeding in MOOC strategy. However, the large-scale development of MOOC platforms has made the curriculum system and knowledge system in MOOC ecosystem lack of organization. This brings a great challenge to implement the personalized learning guidance mechanism. In this project, we aim at helping MOOC learners implement self-directed learning. We propose the problem of personalized learning path recommendation based on learning object network (LON). We construct the high-quality LON by mining the open, multi-source and heterogeneous MOOC educational resource. Based on LON, we aim at making breakthrough on a set of key problems in the research of personalized learning path recommendation in MOOC learning environment. Specifically, we propose four related approaches to tackle these problems, which include semantic mining of learning objects based on crowd wisdom, construction and measurement of LON based on heterogeneous information network, learner semantic modeling based on LON and multiverse context-sensitive learning path recommendation. This project aims at developing feasible theory and framework for personalized learning path recommendation in MOOC environment and advances the research level and innovation ability of our country in the research of MOOC online education.

在推动慕课建设的过程中，能否有效地帮助学习者改善学习管理、规划学习路径、实现自我导向学习，是影响慕课战略能否成功实施的关键因素之一。当前各大慕课平台的规模化发展使得慕课生态呈现出课程体系和知识体系缺乏有序组织的问题，这对慕课个性化学习引导机制的实现带来很大挑战。本项目以指导慕课学习者实现自主导向学习为动机，提出基于学习对象网络的个性化学习路径推荐问题，通过深入分析挖掘慕课学习环境中开放多源异构的学习资源数据构建高质量的学习对象网络，然后基于学习对象网络突破慕课学习环境下个性化学习路径推荐的关键技术。具体研究包括基于群体智慧的学习对象语义挖掘，基于异构信息网络技术的学习对象网络构建和度量，基于学习对象网络的学习者语义建模和基于多元上下文感知的学习路径推荐。本项目旨在建立一套切实可行的慕课学习个性化学习路径推荐的理论和框架，提升我国在慕课在线教育领域的研究水平和自主创新能力。

在推动慕课建设的过程中，能否有效地帮助学习者改善学习管理、规划学习路径、实现自我导向学习，是影响慕课战略能否成功实施的关键因素之一。本项目研究基于学习对象网络的个性化学习路径推荐问题，主要研究成果体现在如下三个方面。一是提出了一个面向在线开放课程领域的学习对象网络构建框架，包括教学资源语义标注、外部知识库语义关联以及课程质量度量等方法，基于这些方法构建了一个大规模课程知识图谱；二是针对在线教育环境中的学习者兴趣建模问题，提出了两种基于深度神经网络的学习者知识状态追踪方法，并设计了一种基于主动学习的学习者兴趣获取框架；三是基于学习对象网络和学习者兴趣建模技术，综合利用课程资源网络结构分析和学习者历史学习数据挖掘等技术，提出了两种基于深度神经网络的在线课程推荐算法并与真实慕课平台集成。