摘要： 提出了一个结合Web文本挖掘的分布式Web使用挖掘模型DWLMST，以及基于该模型的局部浏览兴趣迁移模式更新算法LITP和全局浏览兴趣迁移模式更新算法GITP。利用页面聚类来表示用户兴趣。通过将用户事务中的页面替代为相应的聚类号来得到用户浏览兴趣序列。从用户浏览兴趣序列中分析得到用户浏览兴趣迁移模式。算法较好地解决了Web访问信息的异地存储、实时增长等因素给模式分析过程带来的困难，同时也提高了用户浏览兴趣表示的准确性。

关键词: Web使用模式挖掘, 分布式数据挖掘, Web日志挖掘

Abstract: This paper proposes a distributed Web log mining system model based on Web text(DWLMST). It is presented that the more general incremental updating algorithms of local interest transfer patterns (LITP) and global interest transfer patterns (GITP) in a distributed database system based on DWLMST for discovering the user interest transfer patterns. The algorithms better solve troubles which are made by the real time incremental distributed Web access information and the more communication number.

Key words: Web usage pattern mining, Distributed data mining, Web log mining