The semantics of aggregate queries in data exchange revisited

Phokion G. Kolaitis, Francesca Spezzano

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Defining "good" semantics for non-monotonic queries and for aggregate queries in the context of data exchange has turned out to be a challenging problem for a number of reasons, including the dependence of the semantics of the concrete syntactic representation of the schema mapping at hand. In this paper, we revisit the semantics of aggregate queries in data exchange by introducing the aggregate most-certain answers, a new semantics that is invariant under logical equivalence. Informally, the aggregate most-certain answers are obtained by taking the intersection of the aggregate certain answers over all schema mappings that are logically equivalent to the given schema mapping. Our main technical result is that for schema mappings specified by source-to-target tuple-generating dependencies only (no target constraints), the aggregate most-certain answers w.r.t. a schema mapping coincide with the aggregate certain answers w.r.t. the schema mapping in normal form associated with the given schema mapping. This result provides an intrinsic justification for using schema mappings in normal form and, at the same time, implies that the aggregate most-certain answers are computable in polynomial time. We also consider the semantics of aggregate queries w.r.t. schema mappings whose specification includes target constraints, and discuss some of the delicate issues involved in defining rigorous semantics for such schema mappings.

Original languageEnglish
Title of host publicationScalable Uncertainty Management - 7th International Conference, SUM 2013, Proceedings
PublisherSpringer Verlag
Pages233-246
Number of pages14
ISBN (Print)9783642403804
DOIs
StatePublished - 2013
Event7th International Conference on Scalable Uncertainty Management, SUM 2013 - Washington, DC, United States
Duration: 16 Sep 201318 Sep 2013

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8078 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference7th International Conference on Scalable Uncertainty Management, SUM 2013
Country/TerritoryUnited States
CityWashington, DC
Period16/09/1318/09/13

Fingerprint

Dive into the research topics of 'The semantics of aggregate queries in data exchange revisited'. Together they form a unique fingerprint.

Cite this