2006. Manuscript.
In this paper we present a mechanism for answering queries over relational data sources that are described by means of a conceptual schema, under the assumption that the data in the sources may be incomplete w.r.t. the information in the schema. Following the technique in (Calvanese et al., AAAI05), the query answering process is based on query rewriting, and expands the initial query by taking into account the constraints imposed by the conceptual schema. We provide an expressive schema language which includes features from UML and ER diagrams, as well as ontology languages like Description Logics. Indeed, we extend the framework of (Calvanese et al., AAAI05) to relations of arbitrary arity, allowing also for non-unary keys, provided suitable syntactic restrictions are satisfied. The schema language enables the domain modeller to describe the data sources using the logical schema of the underlying relational databases. This mechanism provides a great flexibility w.r.t. the data sources, making it applicable also to information integration scenarios. Using query rewriting allows the system to overcome incompleteness in the data w.r.t. the constraints imposed by the conceptual schema. Indeed, the system is able to infer additional information that was not stated explicitly in the initial query, and the constraints do not need to be considered by the user formulating the query. An important aspect is that the query rewriting process is independent of the actual data in the information sources, since it uses the information from the conceptual schema only. Moreover, the efficiency of the query answering is guaranteed by the polynomial data complexity of the query answering problem. We provide also an algorithm to verify the consistency between the conceptual schema and the actual data. This consistency check uses SQL queries to verify that constraints in the schema are not violated by the actual database. Our approach has been implemented in a prototype system.
@unpublished{ER-2006,
title = "Answering Queries over Relational Databases using their Conceptual
Schema",
year = "2006",
author = "Diego Calvanese and Mantas Simkus and Sergio Tessaris",
note = "Manuscript",
}