Abstract: Information push service has been widely applied to various e-commerce sites. However, there are some defects when using the two wildly used information filtering techniques in the process of forest products trading. On the base of summing up the features of forest products trading information filtering, the cascade filtering technique is brought up. The technology is based on the different interest of forest products trading users, the different level of users’ needs, the different weight of forest products trading information’s feature item. After getting the vector space model of forest products trading information and analyzing the structure of users’ interest feature vector, the cascade filtering technique is used to exactly meet the user’s needs. It is brought up by strictly filtering the user’s dominant needs firstly, roughly filtering the user’s dominant needs secondly, filtering the user’s implicit needs finally. The experimental results show the cascade filtering technique can efficiently filter the user’s satisfied information.

Key words: information filtering, forest products trading information, information push, information service, recommender system

摘要： 目前信息推送服务广泛应用于各类电子商务网站，然而传统信息过滤技术在林产品贸易信息过滤的过程中，存在着不足。在总结林产品贸易信息过滤的特点的基础上，提出梯级过滤技术。该技术根据林产品贸易用户兴趣的不同、用户需求程度的不同、林产品贸易信息特征项权重的不同，在提出林产品贸易信息特征向量空间模型和分析用户兴趣特征向量结构的基础上，采用先进行用户显性需求的严格过滤，再进行用户显性需求的模糊过滤，再进行用户隐性需求的过滤的梯级过滤方法，从而达到准确满足用户需求的目的。实验结果证明该方法能够有效地过滤出用户满意的信息。

关键词: 信息过滤, 林产品贸易信息, 信息推送, 信息服务, 推荐系统