Topic Knowledge Oriented On-demand Value Services Discovery Approach

In the era of cloud computing and big data,the data more profoundly affected the calculation results than the algorithm.According to the principle of "requirement is the measure of value",the value is the service from the perspective of on-demand service,namely VaaS(Value as a Service).The value can be discovered from data and used by users,which can be understood as finding the value of the service on-demand from data.Faced with the emergence of various types of data resources on the Internet,it becomes an important problem to be solved that how to acquire knowledge to support the value service discovery.With the rapid increase of data size and the complexity of the data structures,it brings great challenges to the value service discovery.At the same time,the user level on the Internet presents a diversified trend,and different levels of users are also showing a trend of individuation and diversification.These conditions have further exacerbated the difficulty of on-demand value service discovery.Aiming at the above problems,the key problem to be solved is:How to acquire knowledge from the emerging data resources to support on-demand value services discovery.The main innovations of this thesis are as follows:(1)This thesis presents an iterative method based on SVM for topic classification,which is used to organize the data resources from the perspective of the topic.The procedure can obtain the corresponding topic knowledge.Firstly the method collects the description of documents from domain data,and then integrate it into a vector form based on the vector space model.Finally the iterative topic classification method classifies the data resources oriented to topic.At the same time,we can obtain the topic knowledge based on the topic vocabulary sorting tables,which lay the foundation for the subsequent sub topic clustering.(2)This thesis presents a clustering method based on LDA oriented to a sub topic by user guidance.The method can organize data resources for more fine-grained from the sub topic perspective,and get more detailed knowledge.On the basis of topic classification,the description of documents in specific topic data set are clustered for sub topic and we can obtain the sub topic cluster set.According to the clustering results,the correlation degree between the data and the sub topic clusters is calculated,and the data sets in the sub topic clusters are ranked according to the correlation degree.At the same time,we can get more detailed topic knowledge based on the feature vocabulary sorting tables,and provide support for the on-demand value service discovery.(3)This thesis presents an on-demand value service discovery method based on RGPS,which is used to locate the value service matching the requirement.This method makes the requirement with the functional similarity calculation from the topic to the sub topic,based on the topic knowledge of classification and clustering.And then through the functional and nonfunctional similarity calculation and combined filtering strategy,the method can search the value service from the sub topic clusters.This method is helpful for users to quickly find the value service to meet their individual requirement.(4)Through the application in the cloud service supermarket,this thesis presents the practical application value of the method.The application in the construction of topic knowledge ontology reflects the good application prospect of the method.The key problem to be solved is:How to acquire knowledge from the emerging data resources to support on-demand value services discovery.On the basis of topic classification and sub topic clustering,this thesis proposes a novel method of progressive on-demand value service discovery.With the combination of topic knowledge and a variety of strategies,the method can step by step locates the value service required by the user.The topic oriented classification and sub topic oriented clustering are the organization and management for data resources from different perspectives,different levels and different granularities.In this process,the topic knowledge is gradually presented,which provides support for the on-demand value service discovery.