The goal of this research project is to study the interplay between constraints and mappings in interactions from XML to XML sources, or XML to relational data sources. The approach consists of developing a language (or languages) that is (are) sufficiently expressive to capture interesting classes of constraints, structure and mappings, and techniques for reasoning about how constraints are translated through these mappings. Using these techniques, algorithms for reasoning about the correctness of mappings with respect to constraints will be developed. Since an XML view also represents a mapping between two different XML sources, one of which is virtual, the related question of how to map an update on an XML view to the underlying data source will also be considered.
The results of this work will provide the ability to detect whether or not semantic conflicts will arise before data exchange actually occurs, thus avoiding time-consuming and unanticipated errors as data loading is performed at the target site. The project is motivated by problems in bioinformatics applications involving gene expression data sharing between projects in the Penn Center Bioinformatics. Since data exchange occurs in many different application domains, for example e-commerce, science, and government, the impact will be broadly applicable to all these areas.
The research performed will be included in an advanced database course taught by the PI, and undergraduate students will be involved in the research through a senior projects course. The results will be broadly disseminated via the project's Web site (www.cis.upenn.edu/~susan/NSF-IDM2005.htm).
