Reconciliation and Merging of Heterogeneous XML Data

With the popularity of information repositories, merging information available at various heterogeneous data sources has become an interesting problem. The data at the various sources might itself be structured, semi-structured or unstructured and the integration process involves reconciliation at various levels. XML has recently emerged as a good candidate in providing a unified view through a common data model. This paper surveys existing approaches to data integration and presents a novel architecture to data integration using XML and web services. This paper is organized as follows: Section 1 presents a classification of data integration systems and outlines reasons to choose XML as a data model in data integration systems. In section 2, we present reconciliation techniques at various levels -- at the level of data model, data schema and data instances. In section 3, we survey existing approaches to data integration. We present and compare eight different integration systems. In section 4, we present our architecture to web services based data integration architecture. Finally, section 5 concludes the paper