基于机器学习的线上实验课程资源挖掘

Resource mining of online experimental courses based on machine learning

为实现线上实验课程资源的个性化推荐,提出基于机器学习的线上实验课程资源挖掘,精准挖掘用户偏好.在PC端、移动端两种登录方式下,线上实验课程资源组织结构的独立性及整体性.采用机器学习方法解析线上实验课程资源文件,通过生成项对、根据潜在函数值输出聚类结果,生成不同类别文件事件模板.通过准确率、F_measure、Rand index指标衡量聚类效果.以Agent作为在线实验课程平台的智能化服务模块,依据用户当下搜索路径,通过协同过滤技术实现用户感兴趣文件事件模板的挖掘,并生成聚集树.以此挖掘用户搜索路径关联规则,根据推荐度参数获取线上实验课程资源推荐集.实验结果表明:该方法可挖掘出与用户偏好90%相关的课程资源,且具有较高的比率,列表长度为14、好友数量为15~21个时的新用户挖掘效果最显著.

In order to realize the personalized recommendation of online experimental course resources,a resource mining of online experimental courses based on machine learning is proposed to accurately mine user preferences.According to the independence and integrity of the online experimental course resource organization structure under the two login modes of PC and mobile,the online experimental course resource files are analyzed by machine learning method,and the event templates of different types of files are generated by generating item pairs and outputting clustering results according to potential function values_Measure and Rand index measure the clustering effect.The agent is used as the intelligent service module of the online experimental course platform.According to the user’s current search path,the event template of files of interest to users is mined through collaborative filtering technology,and an aggregation tree is generated to mine the association rules of user search path,and obtain the online experimental course resource recommendation set according to the recommendation parameters.The experimental results show that this method can mine 90%of the course resources related to user preferences,and has a high rate.The new user mining effect is the most significant when the list length is 14 and the number of friends is 15~21.