Mario Vacca

A Normalization Framework for Multimedia Databases

We present a normalization framework for the design of multimedia database schemas with reduced manipulation anomalies. To this end, we introduce new extended dependencies. Such dependencies are based on distance functions that are used for detecting semantic relationships between complex data types. Based upon these new dependencies, we have defined five multimedia normal forms. Finally, we have performed a simulation on a large image data set to analyze the impact of the proposed framework in the context of content-based retrieval applications and in e-learning applications.

A logic framework to support database refactoring

We propose a formal framework for database refactoring, analyzing both the changes to the database schema, and their impact on queries. The framework defines a logic model of changes, and views the database refactoring process as an agent based one. The agent tries to discover and resolve inconsistencies, and it is modeled as a problem solver capable to perform changes triggered upon the detection of database schema anomalies. The framework can be considered a first step towards the automation of the database refactoring process.

A dialogue-based model for the query synchronization problem

The synchronization of queries is one of the schema evolution problems and it calls for the redefinition of those queries becoming undefined after a schema change, in order to keep them still working on the new schema. This problem is particularly difficult for changes that upset the schema, because it could not be possible to rewrite the queries exactly. In this paper we show that the dialogue can play an important role in these cases and we propose a model of dialogue for query synchronization based on the Hintikka interrogative logic.

A MAPREDUCE FRAMEWORK FOR CHANGE PROPAGATION IN GEOGRAPHIC DATABASES

Updating a schema is a very important activity which occurs naturally during the life cycle of database systems, due to different causes. A challenging problem arising when a schema evolves is the change propagation problem, i.e. the updating of the database ground instances to make them consistent with the evolved schema. Spatial datasets, a stored representation of geographical areas, are VLDBs and so the change propagation process, involving an enormous mass of data among geographical distributed nodes, is very expensive and call for efficient processing. Moreover, the problem of designing languages and tools for spatial data sets change propagation is relevant, for the shortage of tools for schema evolution, and, in particular, for the limitations of those for spatial data sets. In this paper, we take in account both efficiency and limitations and we propose an instance update language, based on the efficient and popular Map-Reduce Google programming paradigm, which allows to perform in a parallel way a wide category of schema changes. A system embodying the language has been implementing.