医院网站日志挖掘数据预处理的研究

Data Preprocessing of Hospital Web Log Mining

数据预处理是数据挖掘工作的基础。以医院网站服务器用户访问日志为例,对日志挖掘的概念以及数据预处理过程进行研究。采用数据库技术和Matlab工具消除初始数据集的冗余度、复杂度,以网站页面和用户关联性为度量,得出可直接作为挖掘算法输入的相似度矩阵。重点研究了医院网站日志的用户识别以及相似度矩阵的页面聚类标准化方法。实验结果表明:该方法极大降低了初始日志数据冗余度和维度,提高了预处理结果准确性,增强了数据挖掘算法输入的信息含量。

Data preprocessing is the basis of data mining.This paper took hospital website access records as an example,studied the concept and data preprocessing.It got a similarity matrix as algorithm input by using database technology and Matlab to eliminate data redundancy and complexity and setting up a metric of web page and user relevance.The user identification and page clustering criterion of the similarity matrix is focused.The result greatly reduces the data redundancy and dimension,improves the accuracy of preprocessing and increases the input information content of data mining algorithm.