Research On Online Service Reputation Measurement Method Based On Multi-dimensional Evaluation Information

With the continuous development of the Internet and pervasive computing technologies,various online services have exploded.Online services are widely used due to their convenience,simple operation,and low cost,and they have become the main driving force for the development of today's service industry.However,when faced with a large number of functionally identical services,how to make a reasonable choice,so that the user must not only consider the functional requirements,but also consider the non-functional attributes of the quality of service needs.As an important indicator to measure the quality of service,reputation is one of the main information for users to refer to when selecting a service.However,because of the characteristics of openness,dynamics,and separation of space and time,it cannot be guaranteed that all online publishing services are objective,reasonable,and credible.Therefore,it is very meaningful to study an objective online service reputation measurement method to reflect the true reputation of the service.Although many current online service reputation measurement methods have greatly improved the accuracy and objectivity of reputation measurement,most online service reputation measurement methods are single-dimensional.The reputation measurement is only given for historical transactions.A single evaluation does not distinguish the quality of attributes in each dimension of the service.The evaluation granularity is relatively coarse,and it is difficult to objectively reflect the actual comprehensive performance of the service.Second,the attribute information structure identifiable by the single-dimension reputation measurement model is single,and it lacks adaptability to heterogeneous service environments with different attribute quality descriptions.Therefore,this paper proposes an online service reputation measurement method based on multi-dimensional evaluation information.Based on Support Vector Machine(SVM)theory,this method establishes a SVM multi-classifier model for services.It trains SVM classifiers through semi-supervised machine learning.At the same time,it reduces the attribute characteristics in order to improve the generalization ability of the model.In the dimension,SVM multi-classifier model with higher performance is finally used to classify services,and comprehensive multidimensional evaluation information of reputation metrics is implemented.Through experimental verification,it is proved that the proposed reputation measurement method is applicable to online service reputation measurement.Finally,according to the method proposed in this paper,the multidimensional online service reputation measurement prototype system is designed and implemented.