SRAUT: Service Resource Availability analysis method with User Tolerance concern

How embarrassing it is to give up the excellent service recommended by the system just because it is not available. This is indeed common when people use service recommendation system in practice. In traditional recommendation system research, the main factors are User profile, QoS, SLA, etc. However, in reality, if the recommended resources are not immediately available, users will not hesitate to ignore them. Long waiting time or waiting queue is not very friendly to users. In the long run, if such case always happens, it will affect users’ perception of the recommendation system, even the recommended services. Therefore, this paper proposes a service resource availability analysis method considering user tolerance (SRAUT). The service resource availability will be calculated by comprehensively utilizing the waiting time of the service and the different tolerance of different people to the waiting time. In other words, it varies from person to person. In order to achieve this, we conducted a controlled experiment on a representative population. This experiment helps to obtain the user tolerance representation of different groups by using disordered multinomial classification logistic regression. Finally, the traditional queuing model is improved by using this representation to form a more personalized service resource availability analysis method. The analysis results can be used to improve the traditional recommendation algorithm. In order to prove the effectiveness of this modification, this paper carried out verification experiments in the same controlled experimental population. The experimental results show that the SRAUT can significantly reduce the user loss rate and the invalid waiting time compared with the traditional queuing model without considering the user tolerance.