顾客偏好的动态挖掘算法

基于顾客偏好随时间变化的特性，采用聚类、关联规则等技术，对顾客偏好进行动态挖掘．通过追踪顾客购买序列，最终产生Top N产品推荐，旨在提高推荐系统的推荐质量．然后选取协同过滤算法作对照，并采用MovieLens站点提供的测试数据集．通过对召回率和精度两项指标的分析，表明该动态挖掘算法具有较高的推荐准确度和全面性．

Dynamic Mining Algorithm for Customer Preference

According to the characteristics of customer preference that changes with time, customer preferences are mined dynamically with such technologies as clustering and association rules. Purchase sequences of customers are traced, and Top-N product recommendations are generated to improve the recommending quality of the recommen- dation system. Then collaborative filtering algorithm is chosen as a contrast and the test data set provided by MovieLens web site is adopted. Analysis on recall rate and precision demonstrates that the presented dynamic mining algorithm is of higher recommendation accuracy and comprehensiveness.