基于双层聚类方法的网页推荐模型

Scheme of web page usage prediction based on two-layer clustering method

研究web用户访问模式的聚类问题,提出了双层的用户访问模式的聚类方法.第一层采用简单易实现的LVQ(学习向量量化)神经网络方法对日志中的用户访问模式进行简单聚类,在第二层的聚类中,采用加权的模糊c-均值的方法对第一层的聚类结果进行聚类.最后根据聚类结果产生描述该类用户行为的加权访问模式,并以此作为网页推荐依据.实验结果验证了该算法的有效性和可行性.

Clustering web users＇ access patterns is discussed, and a two-layer clustering approach of user access patterns is proposed in this paper. At the first layer, the learning vector quantization （LVQ） approach is exploited to group the patterns from web logs into some clusters. At the second layer, the weighted fuzzy c-means approach is developed to deal with the clustering results of the first layer. Finally weighted access patterns are created to describe the surfing behaviors of web users in the class. Then the scheme of predicting web page usage could be built. The effectiveness and feasibility of the approach are testified by the experiment results.