基于可变阶模型的Web访问模式挖掘算法

Algorithm for Mining Web Navigation Patterns Based on Selective Models

提出一种基于Markov的可变阶模型,综合了一阶Markov模型和高阶模型的新方法。该方法通过引入精度因子和所逼近模型的阶数,可以在保持与一阶Markov模型相当的较低复杂度的同时,使预测精度能够逼近高阶Markov模型。通过实验数据说明了所提出算法的正确性和有效性。

In this paper, a novel approach combining one-step Markov and k-step Markov model is proposed. Using precision factor and n-gram, the new algorithm can achieve high prediction precision that is close to M-step Markov model, while lower complexity is kept as the same as one-step Markov model.