Abstract: Although sessions reconstruction is an important step in Web log mining, the sessions recognized by existing methods are not accurate. An algorithm resolving this problem is proposed. This paper improves the Timeout method with the algorithm. Finally the algorithm enhancing the quality of Web log sessions is proved by experiments.

Key words: Web log mining, Data preprocessing, Sessions reconstruction

摘要： 会话识别是Web日志挖掘的关键步骤，然而很多方法所得到的会话不够精确。该文对此提出优化算法，并对最常用的Timeout方法识别的会话进行优化，通过实验证明会话质量得到了提高。

关键词: Web日志挖掘, 数据预处理, 会话识别