基于关联规则的个体化推荐在传统商业中的应用

随着互联网与电子商务的发展,个性化网页的设计应运而生.本文从多产品(包括服务)提供商角度出发,以数据挖掘关联规则理论为基础,总结了"个性化推荐"在多产品提供商中的应用.推荐计算过程主要有四步:构建知识集;基于客户已知信息的筛选规则;对每个候选产品计算总评分;选择最终推荐产品.并指出:知识集不仅包括历史数据关联规则集,也包括业务知识集;评分可以根据具体业务选择合适的规则度量(置信度、lift值等);按照某些准则得到各候选产品的总评分;这些准则包括加权平均、最大值、最小值等;根据一些诸如评分大小等标准确定最终推荐产品.从一个例子,本文描述了个性化推荐过程.

Personalized Recommendation Based on Association Rules

Over the last few years personalization became an important business problem in webpage designing and one-to-one marketing.This article summarized the personalized recommendation in multi-product supplying business as viewed from data mining.The steps of personalized recommendation is:(1) to form the repository based on the transaction data and business knowledge,(2) to select the rules from the repository based on the known information of the customer,(3) to compute the total score for each candidate product according to some criteria,(4) to select the final product recommended based on their total score with reference to some standard such as the maximum score,or the most reliable and maximum score.