采用Multi—Agent系统中各Agent成员间相互协作的协调机制,并遵循FIPA协议在JADE平台上实现了消息模式的通信机制,将SQLServer2005作为后台数据库,创建了基于Multi—Agent的超市客户关系管理系统。采用基于Agent的蚁群聚类算法对预测为流失的客户进行客户价值细分,通过细分结果评定客户价值等级,以辅助超市管理者制定营销策略,提高了系统的自治性和智能性,达到在各商家激烈的竞争中保持客户的目的。
The coordination mechanism of Agent members in Multi-Agent system is adopted. The communication mechanism of the message mode is achieved based on the JADE platform, which followed the FIPA protocol. SQL Server 2005 is used as a background database. The supermarket customer relationship management system is established by Multi-Agent technology. The ant colony algorithm based on Agent is used to classify the customers, which are predicted as lost customers. The customer value level is assessed by subdivision result, which assisted supermarket supervisor to constitute sell strategy. The management system has higher autonomy and intelligence to keep existing customer in the fierce competition between supermarket enterprises.