Research On Cloud Service Recommendation Based On An Improved K-prototypes Algorithm

The interaction and penetration of cloud computing,Internet of Things,big data,mobile Internet and other new generation in information technology accelerate the cross-border integration of manufacturing,service,financial,business and other related industries.Many internet companies actively build the ecological system which is supported by cloud platform and is at the core of cloud service,greatly facilitating the publishing of innovative service and personalized options of global users.Cloud service is more significantly different from traditional goods,and quality of service is an important attribute for the evaluation of cloud services.However,the traditional cloud service recommendation method places emphasis on the accuracy,which has restrict the user's the view to choose the service and neglected the real demand of the users to get the recommendation of diversified cloud service.How to design the recommendation method according to user's demand for the diversity has become an urgent problem to be solved.This paper mainly solves the problem of distance calculation in mixed data clustering and diverse recommendation of cloud service.Firstly,with the data explosion caused by the rapid development of the Internet,the mixed type of data occupies more and more important position.How to carry through the work of mixed data clustering needs more scientific and exact method,this paper comes up with an improved K-prototypes algorithm to improves the classification accuracy of the classification attribute in the mixed data,which could make the object be divided into the cluster which it belongs scientifically.The algorithm has confirmed the validity of dividing the mixed data.Secondly,in order to change the situation that people only pay attention to the accuracy of cloud service recommendation,this paper proposes a kind of cloud service recommendation method based on diversity.By setting different parameters,we can get different accuracy and diversity,and this paper also verify that we can enhance the diversity of cloud services recommendation greatly in the case of sacrificing a small amount of recommendation accuracy,which conforms to the real needs of cloud service users.