Relatore: dott. Carlo Sartiani

mercoledì 5 marzo 2008, ore 16:30, aula 39 del Dipartimento di Matematica

Sommario: The astonishing growth of available information, on both the Web and local computers, has significantly redefined the requirements and the needs of modern data management: indeed, the problem of finding the right piece of information at the right time has evolved and shifted from a relatively centralized context to an highly decentralized and distributed one. While traditional data management has focused on data hosted on a single data source (e.g., a relational or an object-oriented database), the fragmentation of information on multiple heterogeneous sources, often autonomous and accessible only through forms or low-level web services, implies the definition and development of new techniques and tools for their access. This paradigm shift significantly impacts on current data integration solutions. In- deed, integrating remote data sources, which offer no guarantees about the stability of their schemas and/or query capabilities, requires the adoption of new tools for moni- toring these sources and for assisting the administrator or the project manager in the maintenance of the integrated system. Gamma2 is a project aiming at the definition of techniques for the maintenance of schema mappings in traditional and decentralized/distributed data integration systems, with the ultimate goal of providing a complete and uniform maintenance platform. While a completely automatic maintenance solution is not feasible, as the problem is considered AI-complete, Gamma2 will study techniques that automize the maintenance operations and that involve the human intervention only when is strictly necessary. In particular, Gamma2 will focus on four main problems:
1. monitoring the query capabilities and the schemas of remote data sources;
2. identifying potentially harmful changes in the query capabilities and/or schemas of remote data sources;
3. validating a mapping against the newly changed schema of a remote source;
4. suggesting the administrator how to adapt a mapping to a new version of a target schema, so to preserve both the correctness and the semantics of the mapping.