Database Management

Research focus

The Peer review has evaluated this group as Excellent

Research in data management at Politecnico di Milano has a long and solid tradition; forefront books on distributed databases, conceptual database design, logical databases, and active databases contributed to shape the foundations of this discipline in the last two decades. In the last four years, work addressed the enhancement of database technology in several directions, including active, temporal, spatial, and mobile/very small databases. In these fields, emphasis was placed on formally defining the new features required by each specific data management extension, and then inferring properties descending from those definitions, which lead to improved system implementations or to better understanding of the system behaviour. While supporting the technological advances in data management is important, it is perhaps even more important to support developers and users in taking full advantages of the technology; therefore, database research in the group has always been characterized by an emphasis on innovative languages, methods, design support environments and tools, which could bridge technology to its use in real-life applications; in the last four years, we also focused on query and mining languages for XML repositories and on supporting genomic data retrieval. - Work in active databases focused on defining the formal properties of active rules, enabling the development of more powerful rule analysis tools, and on defining strategies for improving the performances of rule sets under a characterization of the load due to passive and active computations. - Work in temporal databases focused on the introduction of process modelling aspects within temporal information management, and specifically on the impact of conceptual aspects (temporal constraints for process modelling exceptions that may occur during process execution) upon architectural issues and choices. - Work on mobile and very small databases focused on efficient logical and physical data structures and access methods for improving the performance of data management on small, mobile devices. Data structures specifically designed for EEPROM Flash Memory storage support as well several other technologies and application scenarios (e.g., WSNs). - Work on database design has addressed new methods and tools for context-based view design, to cope with today’s need of being able to select, according to the current context, the interesting portion of data among huge amounts of information. In this framework, a context model and the associated view derivation methods have been produced (this work is published on Information Systems, 2007). - Work on spatial data focused on solving the interoperability problems encountered in building a spatial data infrastructure (SDI), consistent with the INSPIRE directive, and in the development of an integrated interoperability architecture capable of dealing with the semantic mapping and geometric harmonization issues raised by the design of a strongly integrated SDI at regional level. - The contributions to query language design were focused upon XML and how to make XML repositories more usable, both in terms of user interface and of retrieval success. We designed XQBE (XQuery By Example), a visual query language using examples of XML as a paradigm for querying XML repositories. We also defined Fuzzy XQuery, a language for approximate queries to XML repositories, where fuzzyness addresses both values and tagging structure; the language enables the extraction of ranked results from XML repositories whose structure is only partially known, hence is suited to many practical applications. The efficient evaluation of graph-based query languages has been studied using model checkers. - Work on data mining focused on new paradigms (and corresponding execution environments and algorithms) for extracting association rules and sequential patterns from XML repositories, 29 thus enabling classical mining operations for a new and important class of repositories. We also proposed a summarized representation of XML data to be used when fast and approximate answers are sufficient, so as to improve mining efficiency (this work is in print on ACM-TOIS). Additionally, we worked on extracting unexpected patterns (pseudo-constraints) from relational databases. This method reveals properties on database states not declared as constraints, but whose violation instances are interesting facts, hence it considers data mining from a new, fully original perspective (this work is in print on ACM-TODS). - Work on genomic data management has produced methodologies and algorithms to effectively use and mine genomic information in heterogeneous and distributed genomic databases. The work also generated a Web-enabled system allowing scientists to select and evaluate efficiently and dynamically the most relevant functional and phenotypic information supporting knowledge discoveries in different biomedical experiments.

Dipartimento di afferenza

Dipartimento di Elettronica e Informazione (DEI)

Docenti afferenti

Stefano Ceri (full professor)
Giuseppe Pelagatti (full professor)
Fabio A. Schreiber (full professor)
Letizia Tanca (full professor)
Cristiana Bolchini (associate professor)
Pier Luca Lanzi (associate professor)
Mauro Negri (associate professor)
Giuseppe Pozzi (associate professor)
Daniele Braga (assistant professor)
Alessandro Campi (assistant professor)
Marco Masseroli (assistant professor)
Elisa Quintarelli (assistant professor)