ONE-TO-MANY DATA TRANSFORMATIONS - As Relational Operations

Paulo Carreira

2007

Abstract

Transforming data is a fundamental operation in data management activities like data integration, legacy data migration, data cleaning, and extract-transform-load processes for data warehousing. Since data often resides on relational databases, data transformations are often implemented as relational queries that aim at leveraging the optimization capabilities of most RDBMSs. However, due to the limited expressive power of Relational Algebra, several important classes of data transformations cannot be specified as SQL queries. In particular, SQL is unable to express data transformations that require the dynamic creation of several tuples for each tuple of the source relation. This paper proposes to address this class of data transformations, common in data management activities, by extending Relational Algebra with a new relational operator named data mapper. A starting contribution of this work consists of studying the formal aspects of the mapper operator focusing on its formal semantics and expressiveness. A further contribution consists of supporting a cost-based optimization of data transformations expressions combining mappers with standard relational operators. To that end, a set of algebraic rewriting rules and different physical execution algorithms are being developed.

Paper Citation

in Harvard Style

Carreira P. (2007). ONE-TO-MANY DATA TRANSFORMATIONS - As Relational Operations . In Proceedings of the Ninth International Conference on Enterprise Information Systems - Volume 1: ICEIS, ISBN 978-972-8865-88-7, pages 503-507. DOI: 10.5220/0002405105030507

in Bibtex Style

@conference{iceis07,
author={Paulo Carreira},
title={ONE-TO-MANY DATA TRANSFORMATIONS - As Relational Operations},
booktitle={Proceedings of the Ninth International Conference on Enterprise Information Systems - Volume 1: ICEIS,},
year={2007},
pages={503-507},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0002405105030507},
isbn={978-972-8865-88-7},
}

in EndNote Style

TY - CONF
JO - Proceedings of the Ninth International Conference on Enterprise Information Systems - Volume 1: ICEIS,
TI - ONE-TO-MANY DATA TRANSFORMATIONS - As Relational Operations
SN - 978-972-8865-88-7
AU - Carreira P.
PY - 2007
SP - 503
EP - 507
DO - 10.5220/0002405105030507