Research On QoS Prediction Method Of Web Services Based On Integrated Spatio-temporal Information

With the rapid development of web service technology,a large number of Web services with the same or similar functions but obvious differences in non functional attributes(such as QoS,quality of service)have emerged in the network.It is particularly important to recommend web services that meet users’ quality of service needs through QoS prediction.QoS based web service recommendation is a research hotspot in the field of service computing.The main method is to use the user’s historical QoS information to predict the missing QoS data of the target user,and then recommend web services.At present,many classical algorithms use collaborative filtering method to predict the QoS of Web services and recommend services,but these algorithms rarely consider the impact of spatial and temporal factors on the QoS prediction of Web services.In practical applications,the temporal and spatial characteristics of Web services are closely related to their QoS performance;Moreover,the extremely sparse data in practical applications has a significant impact on the prediction efficiency and accuracy of Web services.Aiming at the problems of low prediction accuracy and poor scalability of traditional QoS prediction methods,this paper studies the web service QoS prediction method based on spatial information and temporal information.The research results of this paper are as follows:(1)Aiming at the problem of low accuracy of Web Service QoS prediction in the case of sparse data,a web service QoS prediction algorithm based on matrix filling and clustering is proposed.The initial user web service item matrix is filled by PCA dimensionality reduction technology,which enriches the QoS data information in the matrix and improves the accuracy of the prediction algorithm.(2)Considering the influence of spatial information on Web Service QoS prediction,a web service QoS prediction method based on user network location is proposed.By introducing the user’s network location and adding a regular term to the matrix decomposition model to improve the algorithm,predict the missing QoS value of the target user,and finally recommend a better web service that meets the user’s needs to the target user.The experimental results show that after adding the user’s network location information into the matrix decomposition model,the prediction accuracy of Web Service QoS is improved,and the adverse effects of sparse matrix and poor scalability on the prediction results are alleviated to a certain extent.(3)Considering the influence of time information on Web Service QoS prediction,a web service QoS prediction method based on time information is proposed.By introducing time information,the three-dimensional tensor of "user-web service item-time period" is constructed,and the Bayesian method is used to decompose and reconstruct the three-dimensional tensor,fill in the missing tensor data,then predict the missing QoS value of web service,and finally select the better web service to recommend to the target user.The experimental results show that the tensor decomposition model with time information can deeply mine the potential relationship between data,alleviate the impact of QoS dynamic changes on the accuracy of prediction algorithm in the process of web service invocation,and improve the accuracy and robustness of prediction algorithm.The proposed method is verified by experiments on large-scale real data sets,The results show that compared with the traditional web service QoS prediction methods,the prediction results of the proposed method are better.