100 likes | 295 Views
Models for Information Integration: Tuning Local-as-view into Global-as-view. Andrea Cali Giuseppe De Giacomo Maurizio Lenzerini. Local-as-view (LAV) vs. Global-as-view (GAV). Specifying the mapping between the global schema and the sources
E N D
Models for Information Integration: Tuning Local-as-view into Global-as-view Andrea Cali Giuseppe De Giacomo Maurizio Lenzerini
Local-as-view (LAV) vs. Global-as-view (GAV) • Specifying the mapping between the global schema and the sources • Processing queries expressed on the global schema • Maintenance
Problem • Can we transform a data integration system built with the LAV approach into an equivalent system following a GAV approach? GAV LAV equivalent transformation
Framework for Information Integration • Definition 1. A data integration system I is a triple (G, S, M) • G : Global schema • S : Source schema • M : Mapping between G and S
From LAV to GAV • Definition 2. Let I =<G,S,M> be an integration system, having AG as the alphabet of the global schema. An integration system I’ on S is query equivalent to I if for every query q using only the symbols in AG , and for every source databases D: • Query equivalent transformation (I and I’) • Query alphabet • Query answer
LAV System LAV
Example • LAV System I(G,S,M) • G: cites/2 , sameTopic/2 • S: source1, source2 • M: • p(source1) = {<X,Y>|cites(X,Y) & cites(Y,X)} • p(source2) = {<X,Y>|sameTopic(X,Y)&cites(X,Z)&cites(Y,W)} • GAV System I’(G’,S,M’)
Example (Cont.) • GAV System I’(G’,S,M’) • G’: • Relations: cites/2 , sameTopic/2, sourceImage1/2, sourceImage2/2, sourceImageExp1/2, sourceImageExp2/4 • Inclusion dependencies: • sourceImage1[1,2] sourceImageExp1[1,2] • sourceImage2[1,2] sourceImageExp2[1,2] • sourceImageExp1[1,2] cites[1,2] • sourceImageExp1[2,1] cites[1,2] • sourceImageExp2[1,3] cites[1,2] • sourceImageExp2[2,4] cites[1,2] • sourceImageExp2[1,2] sameTopic[1,2] • M’ • p’(sourceImage1) = {<X,Y>|source1(X,Y)} • p’(sourceImage2) = {<X,Y>|source2(X,Y)}
Conclusion • Study the relative expressive power of LAV and GAV approaches to data integration • Transform a LAV system into a GAV system • The global schema is expressed in the relational model with inclusion dependencies • Queries used are expressed in the language of conjunctive queries