PALADIN: A Pattern-based Approach to Large-scale Dynamic Information Integration

The nascent data grid technology aims to make access to globally distributed structured and semi-structured databases possible. However, only integrated, transparent access to such data sources, abstracting not only from distribution and implementation details, but also from the diverse forms of heterogeneity, allows maximum leverage for the envisioned dynamic forms of inter-organisational cooperation. Traditional integration technologies rely on a human-driven development process and are therefore not applicable to the ad-hoc style of integration required in these scenarios. PALADIN aims to automate integration planning by using patterns that capture expert knowledge in a machine-processable way. A pattern describes a generic problem constellation encountered in information integration and provides a guideline to its solution, described as transformations on a graph-oriented representation of the data source schemas and their data.