Research And Implementation Of Service Organization Management Technology Based On Distributed Clustering

In recent years,with the rapid development of cloud computing,mobile Internet and ubiquitous computing technology in the software industry,"everything is connected" and "everything is service" are gradually mature,the available services in various fields are more and more prosperous,and the interconnection between massive services has formed a service Internet.In this context,the increasing number of services and the diversification of service types and service descriptions bring great challenges to the organization management and retrieval efficiency of services.Service clustering method is an effective means of service organization and management.It can identify and aggregate services with a certain degree of similarity.Through the division of service set and service classification,it can realize service reorganization,so as to narrow the search scope of services and improve the retrieval efficiency.Distributed clustering method uses distributed computing framework to optimize the design of clustering algorithm,which can effectively deal with the challenge of massive service resources.At present,the commonly used service clustering methods are usually based on the service description and other text,and achieve service clustering by determining the text similarity.However,there are the following problems:(1)most of the current service description texts are usually short texts with sparse features,which has a great impact on the accuracy of clustering results.At the same time,most of the existing service clustering assume that the service feature vocabulary is independent of each other,which lacks the consideration of the semantic impact of service texts;(2)clustering usually requires iterative calculation,and a large number of disk I / O will become a constraint The bottleneck of distributed clustering efficiency.In addition,the clustering results have no semantic information,so it is difficult to further realize the organization and management of services based on the clustering results.To solve these problems,this paper proposes a service organization management method based on distributed clustering.(1)This paper proposes a service metadata model based on word vector extension and BTM,which makes the diversified service types and service descriptions have a normalized service metadata model;and the word vector extension and BTM application make the service description have semantic characteristics,and alleviate the problem of sparse service description text features.(2)This paper proposes a spark based FCKM service clustering algorithm.Based on the topic model in contribution(1),K-means algorithm and canopy algorithm are combined to reduce the number of iterations;The algorithm is implemented on the memory based spark framework to avoid a large number of disk I / O and improve the efficiency of service clustering.This paper also proposes a semantic modeling method of service class cluster based on feature extraction,which provides support for optimizing service organization and management.This paper designs an experiment to verify the improvement of the clustering effect and efficiency of the proposed service clustering algorithm,and the effectiveness of the semantic modeling method of the proposed service cluster.(3)On the basis of contribution(1)and contribution(2),this paper builds a set of service organization management system,which realizes the functions of service warehousing,service description processing,service clustering,system management,etc.