Data Integration: The Need and the Challenges and the Approaches

Speaker: Yannis Papakonstantinou

Abstract

Business and scientific applications often require access, filtering, integration and transformation of the data of multiple information sources. We discuss three technologies and corresponding architectures, relevant to the problem of data integration. First we discuss materialized views (data warehouses), where the data of interest are collected in advance, cleaned, and transformed into the format the application needs. Then we discuss application integration technologies that enable one to build workflows that appropriately collect and integrate the data that the application needs. The third discussed technology is query-based mediation, where the application is presented with a queriable integrated view of the sources. Then, we discuss issues in selecting a data model for data integration. In particular, we compare the relational model and XML in the context of data integration applications. Finally, we discuss open challenges in the areas of semantic integration and the Web.