摘要： 对商务信息挖掘理论进行了研究，给出了电子商务运作模式的定义、性质、定理及相关证明；提出了一种用于跟踪电子商务活动的数据模型，在模型中定义了网页监控点与商务指标之间的“点-集”映射关系，建立了商务网站的访问事件、网页中事先设定的监控点，以及商务工作流程之间的联系；根据这些研究结果对关联规则挖掘算法中的发现和剪枝过程进行改进，设计了面向电子商务的增量挖掘算法，并在第三方物流信息系统中实现。实践表明，该算法在频繁变化的数据集中的挖掘效率大大高于传统的非增量挖掘算法，基于智能Agent技术的商务信息挖掘模型有效地提高了电子商务的实时跟踪与分析能力。

关键词: 增量挖掘, 电子商务, 关联规则

Abstract: Commercial data mining theories are studied. Web based E-commerce operation pattern is defined and its properties are provided. A mapping theorem is proved, in which a “point-to-set” mapping between Web page links and commerce indexes is presented, establishing the relations between the client browsing patterns, monitoring points in Web pages and commerce process. A new data model is proposed to record business activities. And a new incremental mining algorithm is provided to mine associations between sets of commerce indexes in amazingly increasing data set, by improving discovering and pruning part of present association rule mining algorithm. The method is implemented on third part Web based logistics information system. Experiments show that the algorithm is efficient for data set changed frequently. Based on agent technology, ability of tracking and analyzing commerce information in real time is promoted effectively.

Key words: Incremental mining, E-commerce, Association rule