Service Oriented Heterogeneous Data Integration

With the development of internet and web application, data resources accessed by many distributed applications changed a lot, from one-type, centralized, static to heterogeneous, distributed, and dynamic. For these new web applications, how to share data resources is a hard problem of software, because traditional federated database and distributed database manage system are not able to solve this problem. Service oriented Architecture (SOA) is a new system-design architecture, it deploys, integrate and use loose coupled large element application components in internet, business processes are packaged as services which are implemented using different technologies and can communicate with each other. In this paper we propose service oriented heterogeneous data integration referencing ideas of SOA, first, we divide data integration system into different function components, and then encapsulate these components into services which have a standard for interoperation, so they can request each other no matter what language and technology they are implemented using, at last, we could integrate several services to create a data integration task. The goal of service oriented heterogeneous data integration is to provide a unified data access interface and there is no need to change original data resources' structure, data users could create data integration task using these interfaces according application requirements.In this paper we study research results from home and aboard in the filed of service and data integration, and introduce a concept data model which can describe the structure and semantic information of data resources, then design a dynamically extensive data integration architecture which is service oriented and able to realize data integration effectively. First, we introduce OGSADAI, ontology, web service, characteristics of distributed data and mapping methods of data scheme (GAV and LAV); secondly, we analyze several data models, then propose a data model which is named 'ontology based data model', and can describe objects and their relations in our real world. We can create data model for many kinds of data resources, and this model can meet the requirements of many kinds of new applications. Based upon ontology based data model we introduce a service oriented heterogeneous data integration architecture, and describe several main services, which include data access service (DAS), ontology service (OS), data integration service (DIS) and so on. DAS encapsulates data resources, which are available to DIS. DAS unifies the methods of data access, so DIS could access data without knowing the types of data resources, but we still need the addresses of DASs and data schemes of data resources. In order to simplify the process of data access for end-user, we employ ontology techniques to describe semantic information of data resources, the relations between schemes described using object ontology model, and so, users could access data with knowing any detail of the data resources.We are able to configure the services of our system to meet users' new requirements, and provide a transparent access to all data resources, which are concealed and integrated. In the end, we introduce how we have applied service oriented data integration to SDMSP (Shandong Manufacture Service Platform), and summarize the innovations in theory and practice aspects.