摘要： 基于False-Positive方法挖掘Web点击流中的频繁访问序列时通过相关比率ρ控制其内存消耗和挖掘精度，两者之间会因ρ产生冲突。针对该问题提出一种基于False-Negative方法和时间敏感滑动窗的算法FTS-Stream，该算法利用2个边界参数约束ρ，采用2个边界的加权调和平均数替代ρ。实验证明该算法相对于同类方法有较好的性能。

关键词: 频繁访问序列, 加权调和平均数, 调节因子

Abstract: When using false-positive approach to mine Frequent Traversal Sequence(FTS) in Web clickstreams, it utilizes relaxation ratio ρ to control memory consumption and precision of mining result. However, such approaches may lead to a conflict between precision and memory consumption because of using ρ. This paper proposes FTS-Stream algorithm based on false-negative approach and time-sensitive sliding window to solve the problem. FTS-Stream uses two bounds to constrain ρ, and adopts a Weighted Harmonic Average(WHA) of the two bounds to replace ρ. Experiments show that the algorithm performs better than many established FTS mining methods.

Key words: Frequent Traversal Sequence(FTS), Weighted Harmonic Average(WHA), regulatory factor

中图分类号: 

• TP311