Micro-modeling And Management-optimization Of Internet Online Chatting Service Systems

With the development of e-commerce,the size and quantity of e-commerce platforms are increasing these days.And using the online chatting service systems to serve the customers is a very important part.Compared to telephone or email services,online chatting service systems has its unique advantages,i.e.,the screenshot function and multitask ability.For customers,besides the difference of commodity themselves,the service efficiency and attitude of the platforms have also become the factors that customers account for.However,because of the random arrivals of customers,waiting in the queue is inevitable.In case increasing the servers blindly,there may be a waste of resources,but in case there is less servers than needed,customers may have bad impression on this platform and company.As a result,it’s of vital significance to model and optimize the online chatting service systems.In order to improve the service quality of online chatting systems provided by various e-commerce platforms,this paper constructed a queueing network model to investigate the micro-level evolution of system states,i.e.,the multiple chatting rounds between customers and servers,which was called S model.The abandonment behavior of customers was also considered,not only during their waiting in the queue before getting served,but also between their services after the first-round service.Due to the complexity of the S model,another two approximated queueing models were then proposed,called R model and P model,to bound the original model by solving the QBD equations.In addition,this paper also constructed B model,which assumed that the difference of the service quantity between two of any server during any time is less than 1,as the approximated simulation model of S model.In additional to the analytical results,this paper also tested the model by estimating essential parameters from a real dataset obtained from an anonymous e-commerce company and building stochastic simulation models.The numerical results provided some managerial insights as operating such service systems,such as reducing chatting rounds was better than increasing total service rate in reducing waiting time;increasing M could reduce external waiting time but increase internal waiting time and churn rate;increasing N could reduce internal and external waiting time simultaneously and was better than M.