A Research On Cloud Service Recommendation Based On Performance Awareness

With the rapid development of cloud services,the number of cloud services on the Internet keeps growing.On the one hand,those cloud services,with similar functions but different price,have expand the scope of selection for users.But on the other hand,they also make the users difficult to choose the most suitable one in the face of a wide variety of cloud services.Therefore,the key to the development of cloud service technology is to accurately satisfy the user's needs and to meet the user's preferences.Most of the current cloud service recommendation methods are similar to the common Web service recommendation method,that is,to be predicted according to the service registration information and calculated by collaborative filtering.This cloud service recommendation method has the following shortcomings: first,it can not accurately perceive the performance of cloud services which are just the key concern of the users.Second,the service recommendation can not be calculated according to the user performance preference.To solve the above problems,this paper has done the following work:1.A cloud service recommendation framework based on performance awareness was proposed.The framework was composed of three parts: cloud service resource acquisition module,cloud service performance awareness module and cloud service recommendation module.Cloud service resource acquisition module was used to retrieve the current cloud service system resources;cloud service performance awareness module used optimized BP neural network to perceive cloud service;cloud service recommendation module used AHP calculation module,combined with the performance of sensing results and user preference,to calculate the performance of cloud services.2.A new approach to perceive cloud service performance was proposed.The method used IALR algorithm to optimize the neural network,adopted different network learning rate in different weight coefficient of nodes,and improved the convergence speed through the flatarea.Thus,the speed and accuracy of neural network in the cloud service performance awareness had been improved.Preliminary experiments showed that the framework was faster and more accurate in cloud service performance awareness,and can perform service recommendation calculation based on users' preferences.In this way,it could recommend suitable cloud service to users.