Web日志挖掘数据预处理优化

摘要

In the web log mining, data preprocessing technique is crucial ,which is the basis of the work and will directly influence the later stages.In this paper, choose a inert sod that establish a user access tree to identify the user, based on the existing pretreatment methods, which can take on by the lack of site topology in the case of the log processing.Also give an improved method of transaction processing,which can remove unnecessary transaction derectly without additional steps, thereby reducing the pretreatment results on the impact of data mining to be better applied to the model analysis,to induce the correction on the late mining.%在web日志挖掘中,数据预处理技术是各项工作的基础,此项内容将直接影响到后面的各个阶段,因此,也就显得至关重要.在已有的预处理方法基础上,选用了一种建立用户访问树的方法对用户进行识别,并且可以在缺乏网站拓扑结构的情况下对日志进行处理.提出了一种改进的事务处理方法,这可以在不额外增加步骤的基础上,能直接去除无用的事务,从而减少预处理结果对数据挖掘的影响,使得到的事务能更好地应用到模式分析当中,大大增强了后期挖掘的正确性.