Information integration typically requires the construction of complex artifacts like federated databases, ETL scripts, data warehouses, applications for accessing multiple data sources, and applications that ingest or publish XML. For many companies, it is one of the most complicated IT tasks they face today. To reduce the overall cost, intelligent tools are needed to simplify this difficult task.

Clio is a semi-automatic tool for schema mapping and data integration developed at IBM Almaden Research Center over the past few years. It takes source and target schemas as input, which may describe relational or XML data models. Via a graphical SchemaViewer, a user can then interactively specify attribute correspondences between the source and target schemas. An AttributeMatcher component helps suggest such correspondences, based on the similarity of both attribute names and attribute values. Once the user has specified correspondences, Clio generates SQL, SQL/XML, XQuery or XSLT on the fly to implement the specified transformation, which is guaranteed to produce output data that conforms to the target schema. In this talk, we will first describe and demonstrate some basic features of Clio. In particular, we will describe the abstracted problems and the algorithms behind the AttributeMatcher component. Then, we will describe additional research problems abstracted from the area of schema mapping and information integration, with an emphasis on graph algorithms and issues on scalability and parallelism.