当今社会，学习资源呈现海量化积聚，但是没有与之相匹配的学习资源高效检索与利用方法，基于关键词的检索不能深入到资源的内容层面，不能满足用户最直接的需要；与此同时智能辅助学习系统、智能辅助教学系统的高速发展也需要具有教育意义的学习资源单元来支撑，这样才能实现系统的适应性和个性化服务；学习环境也在高速发展的E-learning技术和Web语义技术的催动下将要发生巨大变革，构建跨系统平台、跨组织的海量知识共享环境成为一种发展趋势。因此，采用自动化或者半自动化的方法来生成元数据并实现学习对象的自动化生成为一种必要的手段。　　基于上述问题，本研究小组深入分析了学习对象自动生成领域的现状，考虑到学习对象逐渐向知识型转变的新趋势，提出了一个面向服务的学习对象元数据生成平台框架。本篇论文致力于延续小组的研究工作，进一步完善支持框架运作的本体技术以及针对音视频资源的学习对象生成与标注模块。　　本文进一步拓展了本体构建的关键技术和方法，完善了本体内部知识点之间的关系计算以及知识空间的边界演绎，提出了一套更加准确合适的本体演化算法来计算知识点之间的概念相似度，并成为后面基于本体的学习对象标注算法中分块归并的重要依据，与此同时也完善了本体的编辑和导出接口，方便了一般用户对于本体的编辑和管理。　　此外，论文还提出了一种基于本体的音视频学习对象语义标注算法，它实现了学习资源的对象化生成，并对其完成标注。本算法先对音视频字幕进行固定间距切分，然后根据每一段文本的特征向量模型表征该分块并对该段文本完成标注，再基于时间轴由前往后的逻辑进行第一轮归并，然后再根据整个文档的标注结果以及各个分块之间的标注结果关联强弱进行第二次归并，最终实现对于各个分块的标注结果。本方法充分利用了教学视频本身授课大纲的结构性，在没有视频底层特征信息支撑的情况下，从文本分析的方法出发实现音视频的切分。通过实验分析，本方法的效果良好，为音视频学习对象的多层次、多粒度重用提供了新方向。

In today's society, Accumulation of learning resources is enormous, but there is no efficient method of retrieval and use about learning resources.Keyword-based search can not go deep into the content level of resources, this does not meet the most immediate needs of the user; At the same time the rapid development of smart-assisted learning systems, intelligent CAI systems also need to have educational learning resources to support,so that intelligent systems can have flexibility and personalized service; learning environment is changing, with the rapid development of E-learning technology and Semantic Web technologies,building cross-platform, cross-organizational knowledge sharing environment become a huge trend.Therefore, the use of automated or semi-automated methods to generate and assisted learning object metadata generation has become a necessary means.　　Based on the above problems, the research team analyzes the current situation in the field about learning objects automatic generation.Taking into account the new trends in learning object gradually shift to a knowledge-based learning objects, we propose a service-oriented learning object metadata generation platform framework. It will improve the functioning of ontology technology and audio and video resources for learning object annotation module.　　In addition, the paper also presents a ontology-based learning object semantic annotation algorithm for audio and video.It is used to generate the object of learning resources and annotations. The first step is to cut the subtitle into fixed length blocks, then system extract the feature vector of each block, system begins the first round of the merger base on timeline,merger adjacent block marked with the same.Then based on the results of the entire document annotation and the strength between ontology association for each block merge again.Final system labeles for each sub-block.This method makes full use of instructional video itself structured syllabus.Without underlying audio and video features supporting, systems departure from the method of text segmentation to finsh the audio and video segmentation. Through experimental analysis, the effect of the method is good, to achieve audio and video learning objects multi-level,multi-granularity reuse.