BDA 2019, 35ème Conférence sur la Gestion de Données : Principes, Technologies et Applications, 15-18 October 2019, Lyon, France
The manual specification of transformations between heterogeneous schemas is a key activity in data integration. While there are tools exposing correspondences between elements of schemas, precise schema mappings still need to be manually specified every
time, even if the two schemas at hand are similar to others that have already been mapped. In fact, schema mappings are defined over the properties of the original schemas and cannot be reused on new ones. We tackle the transformation reuse problem by generalizing schema mappings as meta-mappings, a novel formalism describing
transformations between generic data structures called metaschemas. We introduce techniques to infer meta-mappings from schema mappings. Once inferred, we organize meta-mappings into a repository, which can be efficiently searched to identify reusable
transformations when a new pair of schemas is given.We report effectiveness
results from an experimental evaluation over real-world scenarios and show that our system can infer, store, and search millions of meta-mappings in seconds.
Type:
Poster / Demo
City:
Lyon
Date:
2019-10-15
Department:
Data Science
Eurecom Ref:
6043
Copyright:
© EURECOM. Personal use of this material is permitted. The definitive version of this paper was published in BDA 2019, 35ème Conférence sur la Gestion de Données : Principes, Technologies et Applications, 15-18 October 2019, Lyon, France and is available at :
See also:
