简介：Asystematic,efficientcompilationmethodforqueryevaluationofDeductiveDatabases(DeDB)isproposedinthispaper.Inordertoeliminateredundancyandtominimizethepotentiallyrelevantfacts,whicharetwokeyissuestotheefficiencyofaDeDB,thecompilationprocessisdecomposedintotwophases.Thefirstisthepre-compilationphase,whichisresponsiblefortheminimizationofthepotentiallyrelevantfacts.Thesecond,whichwerefertoasthegeneralcompilationphase,isresponsiblefortheeliminationofredundancy.Therule/goalgraphdevisedbyJ.D.Ullmanisappropriatelyextendedandusedasauniformformalism.Twogeneralalgorithmscorrespondingtothetwophasesrespectivelyaredescribedintuitivelyandformally.

简介：

简介：Aconceptualleveldatabaselanguagefortheentityrelationship(ER)modelimplicitlycontainsintegritiesbasictoERconceptsandspecialretrievalsemanticsforinheritancesofattributesandrelationships.Prolog,whichbelongstothelogicalandphysicallevel,cannotbeusedasafoundationtodirectlydefinethedatabaselanguage.ItisshownhowPrologcanbeenhancedtounderstandtheconceptsofentities,relationships,attributesandis-arelationships.TheenhancedPrologisthenusedasafoundationtodefinethesemanticsofadatabasequerylanguagefortheERmodel.Thethreebasicfunctionsofmodelspecification,updatesandretrievalsaredefined.

简介：Approximatequeryprocessinghasemergedasanapproachtodealingwiththehugedatavolumeandcomplexqueriesintheenvironmentofdatawarehouse.Inthispaper,wepresentanovelmethodthatprovidesapproximateanswerstoOLAPqueries.Ourmethodisbasedonbuildingacompressed(approximate)datacubebyaclusteringtechniqueandusingthiscompresseddatacubetoprovideanswerstoqueriesdirectly,soitimprovestheperformanceofthequeries.WealsoprovidethealgorithmoftheOLAPqueriesandtheconfidenceintervalsofqueryresults.AnextensiveexperimentalstudywiththeOLAPcouncilbenchmarkshowstheeffectivenessandscalabilityofourcluster-basedapproachcomparedtosampling.

简介：

简介：一线性(q，，，m(n))局部地可译码的代码(LDC)C：\mathbbF{\mathbbF}n\mathbbF{\mathbbF}m(n)是从向量空间\mathbbF的线性转变{\mathbbF}n到空间\mathbbF{\mathbbF}每消息标志xi能与概率为被恢复的m(n)至少\frac1|\mathbbF|+e\frac{1}{{\left|\mathbb{F}\right|}}从由查询仅仅q的一个使随机化的算法的C(x)的+\varepsilonC放(x)，就算直到C的m(n)位置(x)被贿赂。在Dvir的一个最近的工作，为线性LDC降低界限的作者表演能为算术电路暗示更低的界限。他建议那证明界限更低因为在建筑群或真实的地上的LDC是为接近他的之一的一个好起点推测。我们的主要结果是m(n)=(n2)为在任何东西上的线性3质问LDC的更低的界限，可能无限，地。常数在(