基于相似模式聚类的电子商务网站个性化推荐系统研究

保证个性化推荐系统产生高质量的推荐结果的重要因素是:系统必须要确定访问者在访问行为的相似程度,从而能预测访问者的访问和购买兴趣。实现此功能的关键技术是计算访问者对象在整个或者部分属性空间的相似距离,从而得到访问行为的相似程度。该文首先分析了目前在推荐系统中常用的用于计算访问行为相似程度的距离函数,发现它们是测定访问者对象在所有测试属性空间上的平均测定,而在属性集的子维空间上的相似模式并没有有效地挖掘出来。然后提出一种新的基于相似模式聚类算法的电子商务个性化推荐系统,综合考虑可供挖掘的数据源(如:网站内容,网站的超链接结构,顾客访问网站的行为,以及商业的实际购买情况,顾客的身份数据等)获取用户访问电子商务网站的访问页面序列,构建较高购买者的顾客行为的矩阵模型,高效地得到访问者对象在整个或者部分属性空间的相似访问行为,然后通过挖掘潜在购买者与较高购买者的相似模式特征,帮助顾客发现他所希望购买的产品信息,用于提高实际购买量,实验数据表明,该系统高效并可广泛使用。

Research on Personalized Recommender Systems Based on Similar Pattern Clustering for E-Commerce

In the applications of recommendation systems in the E-commerce,sets,customers /clients with similar behav-ior need to be indentified so that we can predict customers' interest and make quality recommendation.The key technol-ogy to produce high quality recommendation is to define similarity among different objects by distances over either all or only a subset of the dimensions.Some well-known distance functions are not always capture correlations among the object.The paper proposes a novel similar pattern clustering algorithm that can discover the similar pattern that exhibit a coherent pattern on a subset of dimensions.Based on similar pattern clustering,we give a Personalized Recommender Systems for E-Commerce,the importance goals of recommender system for E-commerce is to increase sales of existing produces by matching customers to the products that will be most likely to purchase.When Visitors navigate though a company`s web site,the similarities in navigational behavior of top-selling visitors can help to increase turnover.New recommender system technologies can quickly produce high quality recommendations.It shows that the system is high ef-ficiency and can be widely used,or is also a suitable technique for recommender system on E-commerce site.