150 likes | 245 Views
ConQuer: Efficient Management of Inconsistent Databases. Presented by: Ariel Fuxman (Univ. of Toronto) Joint work with: Renée J. Miller (Univ of Toronto) Diego Fuxman (Univ. Nacional del Sur). ConQuer. A system designed to answer SQL queries over inconsistent databases. Name.
E N D
ConQuer: Efficient Management of Inconsistent Databases Presented by: Ariel Fuxman (Univ. of Toronto) Joint work with: Renée J. Miller (Univ of Toronto) Diego Fuxman (Univ. Nacional del Sur)
ConQuer • A system designed to answer SQL queries over inconsistent databases Name Income Peter 40K Peter 200K INCONSISTENT DATABASE Paul 400K Mary 110K Mary 130K nameshould be thekey Ariel Fuxman, Diego Fuxman, Renée J. Miller
One Application Customer Relationship Management (CRM) Sales Integrated Customer Database Shipping Customer Support Web Forms Demographic Data Ariel Fuxman, Diego Fuxman, Renée J. Miller
Disagreement Between Sources • Which tuple for Peter should we delete? • Removing both tuples loses consistent information • Deciding the correct income may require human intervention name address … income Peter 276 College Street …. 40K sales Paul 100 Bloor Street … 400K Mary 20 Union Street … 110K name address … income Peter 276 College Street …. 200K web Paul 100 Bloor Street … 400K Mary 20 Union Street … 130K Ariel Fuxman, Diego Fuxman, Renée J. Miller
Inconsistent Integrated Database Transfer all conflicting tuples to the integrated database Integrated Database Sales INCONSISTENT DATABASE Web Ariel Fuxman, Diego Fuxman, Renée J. Miller
Query Answering Offering a Platinum credit card… q=“Get customers who make more than 100K” Peter,Paul,Mary Peter should NOT be offered a Platinum card!! name income Peter 40K sales Peter 200K web Paul 400K sales/web Mary 110K sales Mary 130K web Ariel Fuxman, Diego Fuxman, Renée J. Miller
Semantics of Query Answering • Get customers who possibly make more than 100K • Peter, Paul, Mary • Get customers who certainly make more than 100K • Paul, Mary CONSISTENT ANSWER [Arenas et al. 99] Ariel Fuxman, Diego Fuxman, Renée J. Miller
Repairs Repairs Inconsistent database custid income 40K sales Peter Peter 200K web Paul 400K sales/web Mary 110K sales Mary 130K web Key: custid Ariel Fuxman, Diego Fuxman, Renée J. Miller
Consistent Query Answers q=“Get customers who make more than 100K” Repairs q Paul Mary CONSISTENT ANSWER= {Paul,Mary} CONSISTENT ANSWERSAnswers obtained no matter which repair we choose q Paul Mary Peter q Paul Mary Peter q Paul Mary Ariel Fuxman, Diego Fuxman, Renée J. Miller
Problem Potentially HUGE number of repairs! Ariel Fuxman, Diego Fuxman, Renée J. Miller
ConQuer • ConQuer is a system designed to compute consistent answers efficiently • avoids explicit construction of repairs • reuses commercial database technology Ariel Fuxman, Diego Fuxman, Renée J. Miller
ConQuer’s Solution Query q Keys Consistent answer to q ConQuer’s Rewriting Algorithm [ICDT 05] [SIGMOD 05] Commercial database engine Inconsistent database Rewritten Q* Ariel Fuxman, Diego Fuxman, Renée J. Miller
Contributions • Rewriting algorithm • From a large class of SPJ SQL queries • Into SQL queries • Rewriting for queries with grouping and aggregation • Optimized rewriting • Exploits precomputed information, if available • Experimental evaluation • Large databases • TPC-H queries Ariel Fuxman, Diego Fuxman, Renée J. Miller
Demo • Present a case study of an inconsistent database about airports and cities • Explain the automatically generated rewritings • Deal with Select-Project-Join queries with grouping and aggregation Ariel Fuxman, Diego Fuxman, Renée J. Miller
ConQuer papers • A. Fuxman, E. Fazli, and R. J. Miller. ConQuer: Efficient Management of Inconsistent Databases, SIGMOD 2005. • A. Fuxman and R. J. Miller. First-Order Query Rewriting for Inconsistent Databases, ICDT 2005. Ariel Fuxman, Diego Fuxman, Renée J. Miller