随着AI和大数据技术的发展,人机混同客服中心成为当今企业进行客户服务的常见形式,为实现企业经济效益最大化,对智能客服投入和传统人员的配置方式进行合理决策至关重要。文章在考虑不耐烦顾客的双层排队模型基础上,引入柯布–道格拉斯函数对智能技术投入效益进行刻画,分析了客服中心在人工和智能技术共同作用下的顾客平均放弃率与经济效益,从而作出人机资源的最优配置决策,并对顾客平均达到率、顾客平均耐心时间、技术发展水平等参数进行敏感性分析,为企业进行人机资源投入决策提供了管理建议,促进了经济学与排队论的融合。With the development of AI and big data technology, human-machine hybrid customer service centers have become a common form of customer service for today’s enterprises, and in order to maximize the economic benefits of enterprises, it is crucial to make reasonable decisions on intelligent customer service inputs and traditional staffing methods. In this paper, on the basis of considering the two-layer queuing model of impatient customers, the Cobb-Douglas function is introduced to portray the benefits of intelligent technology inputs, and the average customer abandonment rate and economic benefits of the customer service center under the joint action of artificial and intelligent technology are analysed, so as to make the optimal allocation of human-computer resources decision-making, and sensitivity analysis of the parameters of the average customer attainment rate, the average customer patience time, and the level of technological development, etc., is carried out to provide an opportunity for enterprises to make human-computer resources allocation decisions. The analysis provides management suggestions for enterprises to make human-machine resource investment decisions and promotes the integration of economics and queuing theory.
文章在经典优先级模型的基础之上,引入灵活抢占策略,研究不同优先级顾客在该策略下的平均逗留时间、服务人数等服务指标变化情况。研究表明,在本文的应用场景假设和搭建的仿真模型中,按照所设置的仿真参数值可以准确计算出三种负载场景下两类顾客的指标变化情况,并根据抢占阈值的研究确定不同负载场景下的阈值选择。这也有助于企业在实际运营中的策略选择,为其以往的经验化决策提供支持。Based on the classic priority model, this paper introduces a flexible preemption strategy to study the changes in service indicators such as average stay time and number of customers with different priorities under this strategy. The research shows that in the application scenario hypothesis and simulation model built in this paper, the index changes of two types of customers under three load scenarios can be accurately calculated according to the simulation parameter values set, and the threshold selection under different load scenarios can be determined according to the study of preemption threshold. This also helps enterprises in the actual operation of the strategy choice, providing support for their past empirical decisions.
