基于改进遗传算法和DBSCAN聚类的学习数据深度挖掘方法

A deep data mining method for learning data based on improved genetic algorithm and DBSCAN clustering

为了从在线学习大数据中提取有用信息,实现自适应特征提取和聚类,提出了基于改进模糊遗传算法和DBSCAN聚类的细粒度学习数据挖掘方法。通过在信息管理平台中应用数据挖掘技术,将学习表现评估转换为文本分类问题,基于动态数据分析细粒度的知识获取结果。所提改进的遗传算法自动提取出文本中的最优特征集,利用模糊规则关联测试内容与知识点。最后,利用基于密度的聚类算法得到每个知识点的个体和整体测试结果。实验结果表明,所提方法能够自动处理大量数据,全面准确地分析测试结果中不同知识点的掌握程度,有助于信息管理平台数据的二次开发和深入挖掘。

In order to extract useful information from on-line learning big data,achieve adaptive feature extraction and clustering,and improve the granularity of learning data mining,an improved fuzzy genetic algorithm and DBSCAN clustering-based method for data mining is proposed.By transforming learning level evaluation into a text classification problem,data mining techniques are applied in the information management platform to dynamically analyze fine-grained knowledge acquisition results.The proposed genetic algorithm automatically extracts the optimal feature set from the text and associates the content of the test with the corresponding knowledge points using fuzzy rules.Finally,a density-based clustering algorithm is used to obtain global as well as individual testing performance on each knowledge point.Experimental results show that the proposed method can automatically process large amounts of data,comprehensively and accurately analyze the mastery of knowledge points from test results,which is helpful for the secondary development and in-depth mining of the data collected by the information management platform.