数据集成系统中的查询处理和优化.pdf

墼塑塞盛堡堕旦塑垒塑竺望塑垡些垒!璺堕 Abstract The advent of theImemet provides US the access tomany autonomous and heterogeneous information purpose ofinformationintegration is to support seamless access tothese data dealwim source heterogeneity, many syStems use a mediation architecture,in which amediator processes user queries essing SOurce data. This dissertation develops powerful query processing techniques and architectures forinformation thisframework,the core oftheintegration system is aquery processor called amediator thatimplements theintegrated query processing algorithms. nlon characteristic ofsuch heterogeneous datab船e(HDB)is thatthey simultaneously process large numbers ofqueries/ ability toefficiently handlelargevolumes ofsimultaneous queries iscriticalinmany thispaper,we present a multiple query optimizationtechniques to improve the performance ofsuch heavily loaded HDB. Thenovelty andcontributions ofthis PaDercan besummarized asfollowing: ●Through investigating traditionalquery containment concept’S ing. we present pseudo containment concepts that can efficiently reduce back-join. ●We develop amultiple query optimization(MQO)techniques toimprove the performance ofsuchheavily loaded HDB based on thepseudo containment. Thismethod iSbased onpseudo containment algorithm Can handle relatively smaller sets ofqueries,the finalstep inoptimizing the performance ofsuch systems is togroup alarge number ofqueries into classesofmanageable size. ●Additionally,we present a wrapper implementation method forrapidly building weknown,every datasource need aspecificwrapper, but arelatively smallpart ofthe code dealswiththespecific access detailsof on thisobservation,we use schema elementmapping table and datatypemapping table tOspecify difference among dataSOurCes· Keywords: DataIntegration,Query Containment,Multiple Query optimizations, Pseudo Containment 2 望塑!!!堡竺±竺壅塑丝些塑垡垡!!童第