Abstract
Under the umbrella of database transformations lies a multitude of various operations pertaining to and characterising several problems of database management systems. Be it the preparation of data via data cleaning operations through formats, the restructuring of a database schema prior to its integration with another or the normalization of a legacy database as part of a broader database engineering process, database transformations are ubiquitous in the realm of database theory. A desirable property found across many of those scenarios is a guarantee that the trans formation applied doesn’t result in the loss of information. In the literature related to the preservation of information capacity, this is done by ensuring the equivalence of the source database schema with its counterpart resulting from a given transformation. In recent years, the Transformation Pattern formalism was conceived to express this exact process. Inspired by its intuitive portrayal of database operations and by its underlying theory of losslessness, we decided to look at the problematic it originally stemmed from, among others, and consider the creation of methodologies based on transformation patterns. In this thesis, we present in a new light the transformation pattern formalism along with its application to more complex database schemas. We then propose a database normalization methodology based on these patterns, which for any database schema returns its equivalent in the fifth normal form, accompanied by a bijective pair of mappings between the two schemas. With all non-trivial constraints trans formed away, we present various restructuring possibilities as we delve deeper into the transformation pattern catalogue. At the end of this road lies the canonical abstract relational model, our proposal for a conceptual model in relational database semantics. Returning to the practical realm, we propose the Semantic SQL Transducer, a tool that permits the mutual updating of two equivalent databases. Finally, we take a step back and gaze into what future transformation patterns may be.