metric trigonometr prune cluster uncertain data geometr space acceler probabilist frequent itemset mine model dbase approach manag uncertainti xml schema match abstract advanc machin learn technolog schema match result databas schema deriv coma imprecis numer instanc map schema deriv match result paper studi problem manag map heterogen xml schema observ xml schema map high degre overlap propos data structur call block tree captur common map block tree repres map compact manner gener effici support evalu probabilist twig queri ptq return probabl portion xml document match queri pattern user interest answer highest probabl propos top ptq present effici solut challeng tackl effici gener map schema match problem solv exist algorithm show improv perform solut divid andconqu approach extens evalu realist dataset show approach significantli improv effici gener store queri map anonym fuzzi ident dbase encrypt similar search mine uncertain data probabilist guarante oblivi transfer access control realiz disjunct duplic oblivi transfer access control realiz disjunct duplic dolap olap system analyz sequenc data dhomogen gener privaci preserv data publish cluster uncertain data voronoi diagram dtree index locat probabl densiti function pdf show uk mean algorithm generalis mean algorithm handl uncertain object ineffici ineffici fact uk mean comput expect distanc ed object cluster repres arbitrari pdf expect distanc comput numer integr costli oper propos prune techniqu base voronoi diagram reduc number expect distanc calcul techniqu analyt proven effect basic bound box base techniqu previous literatur introduc tree index organis uncertain object reduc prune overhead conduct experi evalu effect techniqu show techniqu addit combin significantli outperform previous method cluster uncertain data voronoi diagram dtree index locat probabl densiti function pdf show uk mean algorithm generalis mean algorithm handl uncertain object ineffici ineffici fact uk mean comput expect distanc ed object cluster repres arbitrari pdf expect distanc comput numer integr costli oper propos prune techniqu base voronoi diagram reduc number expect distanc calcul techniqu analyt proven effect basic bound box base techniqu previous literatur introduc tree index organis uncertain object reduc prune overhead conduct experi evalu effect techniqu show techniqu addit combin significantli outperform previous method effici skylin evalu partial order domain consider bodi work skylin evalu multidimension data total order attribut domain method attribut partial order domain exist work map partial order domain total order adapt algorithm totallyord domain solv problem method stronger notion domin gener fals posit requir expens domin check paper propos method drawback method map partial order total order inspir lattic theorem shelf skylin algorithm techniqu storag index approach inspir column store enabl effici verif pair object incompat demonstr method order magnitud effici previou work scale problem paramet complex partial order proceed acm confer inform knowledg manag xml data integr fragment join abstract studi problem answer xml queri multipl data sourc schema independ scenario xml schema schema map unavail develop fragment join oper gener oper merg xml fragment base overlap compon formal defin oper propos effici algorithm implement defin schema independ queri process multipl data sourc propos framework solv problem provid theoret analysi experiment result show approach effect effici privaci dpreserv cluster high accuraci low time complex abstract paper propos effici solut high accuraci problem privaci preserv cluster problem studi approach data perturb secur multiparti comput research focu data perturb approach propos algorithm linear time complex base cluster perturb data perform studi real dataset uci machin learn repositori show approach reach accuraci lower distort cluster result previou approach naiv bay classif uncertain data dgovern data interoper framework hong kong releas digit strategi serv blueprint hong kong develop inform commun technolog infrastructur govern leadership import compon strategi put forward develop govern realiz stop deliveri public electron servic hong kong govern launch interoper framework govern initi facilit implement cross depart join govern servic compris part definit set recommend technic specif singl point refer depart contractor implement join project framework formul manag xml messag standard data exchang paper discuss part xml schema design guid reallif applic secur knn comput encrypt databas servic provid googl amazon move saa softwar servic busi turn huge infrastructur cloud comput environ aggress recruit busi run applic platform enforc secur privaci servic model protect data run platform tradit encrypt method aim provid unbreak protect adequ support execut applic databas queri encrypt data paper discuss gener problem secur comput encrypt databas propos sconedb secur comput encrypt databas model captur execut secur requir case studi focu problem nearest neighbor knn comput encrypt databas develop asymmetr scalar product preserv encrypt asp preserv special type scalar product aps construct secur scheme support knn comput encrypt data scheme shown resist practic attack background knowledg level overhead cost extens perform studi carri evalu overhead effici scheme scalabl skylin comput object dbase space partit skylin oper return set multi dimension object subset superior object domin oper consid import multi object analysi larg dataset larg number skylin method propos major focus minim cost high dimension space problem easili cpu bound due larg number comput requir compar object current skylin point scan databas base observ propos dynam index techniqu skylin point integr state art sort base skylin algorithm boost comput perform index domin check approach support theoret analysi experi show scale input size dimension unnecessari domin check avoid effici domin check bitwis oper scalabl skylin comput object dbase space partit skylin oper return set multi dimension object subset superior object domin oper consid import multi object analysi larg dataset larg number skylin method propos major focus minim cost high dimension space problem easili cpu bound due larg number comput requir compar object current skylin point scan databas base observ propos dynam index techniqu skylin point integr state art sort base skylin algorithm boost comput perform index domin check approach support theoret analysi experi show scale input size dimension unnecessari domin check avoid effici domin check bitwis oper audit environ outsourc frequent itemset mine find frequent itemset costli task associ rule mine outsourc task servic provid bring benefit data owner cost relief commit storag comput resourc mine result corrupt servic provid honest make mistak mine process ii lazi reduc costli comput return incomplet result iii malici contamin mine result address integr issu outsourc process data owner verifi correct mine result purpos propos develop audit environ consist databas transform method result verif method main compon audit environ artifici itemset plant aip techniqu provid theoret foundat techniqu prove appropri show probabilist guarante correct verif process analyt experiment studi show techniqu effect effici secur integr outsourc data mine cluster uncertain data voronoi diagram locat probabl densiti function pdf show uk mean algorithm generalis mean algorithm handl uncertain object ineffici ineffici fact uk mean comput expect distanc ed object cluster repres arbitrari pdf expect distanc comput numer integr costli oper propos prune techniqu base voronoi diagram reduc number expect distanc calcul techniqu analyt proven effect basic bound box base techniqu previous literatur introduc tree index organis uncertain object reduc prune overhead conduct experi evalu effect techniqu show techniqu addit combin significantli outperform previous method mine prefer superior inferior exampl mine user prefer plai critic role import applic custom relationship manag crm product servic recommend market campaign paper identifi interest practic problem mine user prefer multidimension space user prefer categor attribut unknown superior inferior exampl provid user learn user prefer categor attribut model problem systemat show mine user prefer superior inferior exampl challeng problem great potenti practic knowledg explor systemat attempt tackl problem propos greedi method show method practic real data set synthet data set protect privaci increment mainten distribut associ rule mine olap sequenc data kind real life data exhibit logic order data item sequenti natur tradit onlin analyt process olap system techniqu design sequenc data incap support sequenc data analysi paper propos concept sequenc olap olap short biggest distinct olap tradit olap sequenc character attribut valu constitut item subsequ substr pattern possess paper studi aspect relat sequenc olap concept sequenc cuboid sequenc data cube introduc prototyp olap system built order valid propos concept prototyp support pattern base group aggreg support olap system implement detail prototyp system experiment result present descript schema semant dbase queri answer abstract propos concept descript schema ds unlik ordinari databas schema ds restrict structur underli databas probabilist descript structur answer keyword queri ds improv semant base queri answer result rank schema wikipedia rich repositori inform facil exploit inform limit typic search www search engin googl user inform keyword lack schema formul queri precis hyperlink wikipedia page page categori tag infobox exploit perform sophist search dbpedia commun make tag build databas rdf triplet allow express precis queri form sparql retriev inform filter fals posit microrna candid cluster dbase approach webform rapid applic develop toolkit write ajax web form applic abstract web form commonli captur data web asynchron javascript xml ajax program interact web form creat ajax program complex model view control mvc code separ paper discuss mvc orient web form develop call webform develop simplifi streamlin web form develop ajax introduc script languag call web form applic languag webfal model web form webform hide program complex gener ajax code directli web form model mainten maxim frequent itemset larg databas secur outsourc associ rule mine effici top dk aggreg rank input top queri combin rank set object return object highest combin score aggreg function bring light kei observ impos phase top algorithm base sort access base propos algorithm design minim number object access comput cost memori requir top search monoton aggreg function provid analysi cost show wors baselin random access algorithm term comput access memori requir side contribut perform space analysi memori requir top algorithm perform sort access case requir space exce memori propos disk base variant algorithm propos optim multiwai top join oper advantag evalu tree binari top join oper final defin studi comput top cube implement roll drill oper cube extens experi synthet real data show compar previou techniqu method access fewer object order magnitud faster discoveri period pattern spatiotempor sequenc applic track analyz spatiotempor data movement obei period pattern object follow rout approxim regular time interv peopl wake time follow rout work everydai discoveri hidden period pattern spatiotempor data provid unveil import inform data analyst exist approach discov period pattern focu symbol sequenc method directli appli spatiotempor sequenc fuzzi spatial locat sequenc paper defin problem mine period pattern spatiotempor data propos effect effici algorithm retriev maxim period pattern addit studi interest variant problem retriev period pattern frequent histori continu subinterv problem discoveri period pattern instanc shift distort demonstr mine techniqu adapt variant final present comprehens experiment evalu show effect effici propos techniqu mine micro darrai data order dpreserv submatrix studi problem pattern base subspac cluster unlik tradit cluster method focu group object similar valu set dimens cluster pattern similar find object exhibit coher pattern rise fall subspac applic pattern base subspac cluster includ dna micro arrai data analysi automat recommend system target market system goal devis pattern base cluster method capabl discov pattern shape discov signific pattern argu previou solut pattern base subspac cluster satisfi requir approach extend idea order preserv submatrix opsm devis algorithm mine opsm show opsm gener cover exist pattern base cluster model propos number extens origin opsm model mine period pattern gap requir sequenc studi problem mine frequent occur period pattern gap requir sequenc charact sequenc length pattern length frequent occur pattern probabl observ randomli pick length subsequ exce threshold applic relat bioinformat interest pattern period gap requir charact match subsequ match charact separ gap size show complex mine problem discuss tradit mine algorithm computation infeas propos practic algorithm solv problem studi characterist present case studi appli algorithm dna sequenc discuss interest pattern obtain case studi effici aggreg rank input top queri combin rank set object return object highest combin score aggreg function bring light kei observ impos phase top algorithm base sort access base propos algorithm design minim number object access comput cost memori requir top search adapt algorithm search variant exact score line increment search top join aggreg function provid extens experi synthet real data show compar previou techniqu method access fewer object order magnitud faster discoveri colloc episod spatiotempor data balanc ensembl approach weight classifi text classif input valid semi dsupervis cluster progress skylin web daccess databas sembiospher semant web approach recommend microarrai cluster servic cluster popular method analyz microarrai data larg number cluster algorithm difficult identifi suitabl task difficult locat download instal run algorithm paper describ matchmak system sembiospher solv problem recommend cluster algorithm base minim user requir input data properti ontolog develop owl express ontolog languag describ algorithm perform addit invok machin understand algorithm make recommend algorithm implement group languag run platform geograph distribut site xml base web servic invok standard current cluster servic transform semant web servic biospher system includ varieti algorithm appli microarrai gene express data analysi algorithm incorpor system effort sembiospher system complet cluster ontolog access progress skylin web daccess databas wsipl xml script languag integr web servic data applic advanc knowledg discoveri data mine discoveri extrem low ddimension cluster semi dsupervis project cluster recent studi suggest project cluster extrem low dimension exist real dataset number project cluster algorithm propos past year identifi cluster dimension lower total number dimens commonli found real dataset gene express profil paper propos algorithm accur identifi project cluster relev dimens total number dimens make robust object function combin object cluster dimens select singl optim problem algorithm util domain knowledg form label object label dimens improv cluster accuraci semi supervis project cluster algorithm theoret analysi experiment result show small amount input knowledg possibl cover portion underli class algorithm improv accur detect cluster dimens relev algorithm target set cluster multipl group object mine micro darrai data order dpreserv submatrix studi problem pattern base subspac cluster unlik tradit cluster method focu group object similar valu set dimens cluster pattern similar find object exhibit coher pattern rise fall subspac applic pattern base subspac cluster includ dna micro arrai data analysi automat recommend system target market system goal devis pattern base cluster method capabl discov pattern shape discov signific pattern argu previou solut pattern base subspac cluster satisfi requir approach extend idea order preserv submatrix opsm devis algorithm mine opsm show opsm gener cover exist pattern base cluster model propos number extens origin opsm model mine frequent spatio dtempor sequenti pattern applic track movement mobil object repres sequenc timestamp locat spatio tempor seri studi problem discov sequenti pattern rout frequent object sequenti pattern mine algorithm transact data directli applic set challeng address fuzzi locat pattern ii identif explicit pattern instanc paper defin pattern element spatial region frequent line segment method transform origin sequenc list sequenc segment detect frequent region heurist propos algorithm find pattern emploi newli propos substr tree structur improv apriori techniqu perform evalu demonstr effect effici approach cross tabl cube mine iceberg cube data warehous mine period pattern gap requir sequenc studi problem mine frequent occur period pattern gap requir sequenc charact sequenc length pattern length frequent occur pattern probabl observ randomli pick length subsequ exce threshold applic relat bioinformat interest pattern period gap requir charact match subsequ match charact separ gap size show complex mine problem discuss tradit mine algorithm computation infeas propos practic algorithm solv problem studi characterist present case studi appli algorithm dna sequenc discuss interest pattern obtain case studi clustag hierarch cluster graph method select tag snp effici algorithm mine increment updat maxim frequent sequenc index structur pattern xml queri process abstract queri semistructur data hard process due complex natur data call special techniqu exist path base index queri process algorithm effici search complex structur simpl path queri high select introduc definit minim infrequ structur mi structur exist data frequent respect support threshold substructur frequent index occurr mi effici locat high select substructur queri improv search perform significantli effici data mine algorithm propos find minim infrequ structur occurr xml data index lightweight data structur fast filter step queri evalu valid effici applic method experiment synthet real data index term queri process xml xsl rdf mine method algorithm document index fast exact warp time seri adapt segment approxim abstract similar search core modul data analysi task includ search classif cluster time seri data dynam time warp dtw proven effect similar measur minim effect shift distort time quadrat cost dtw comput length match sequenc make direct applic databas long time seri expens propos techniqu decompos sequenc number segment cheap approxim thereof comput fast lower bound warp distanc present progress tighter bound reli exist warp constraint final develop index multi step techniqu propos bound perform level filter effici process similar queri experiment studi suggest method consist outperform state art method dtw similar search identifi project cluster gene express profil microarrai gene express data cluster hide subspac set regul gene similar express pattern subset sampl regul factor present express pattern dissimilar measur full input space tradit cluster algorithm make similar measur fail identifi cluster recent year number algorithm propos identifi kind project cluster reli critic paramet proper valu hard user determin paper algorithm dynam adjust intern threshold propos low depend user paramet allow user input domain knowledg experiment result show algorithm capabl identifi interest project cluster sf dtree effici flexibl structur estim select simpl path express statist accuraci guarante process ad dhoc join mobil devic abstract mobil devic capabl retriev process data remot databas wireless data transmiss environ user typic charg size transfer data amount time stai connect propos algorithm join inform collabor remot databas mobil devic method minim data transfer join process limit mobil devic experiment result show approach perform join process mobil devic effect mine index queri histor spatiotempor data abstract larg spatio tempor databas event event consist field event id time locat event type mine spatio tempor sequenti pattern identifi signific event type sequenc spatio tempor sequenti pattern crucial investig spatial tempor evolut phenomena applic domain recent research literatur explor sequenti pattern transact data trajectori analysi move object method directli appli mine sequenti pattern larg number spatio tempor event major research challeng remain definit signific measur spatio tempor sequenti pattern avoid spuriou algorithm design signific measur guarante downward closur properti paper propos sequenc index signific measur spatio tempor sequenti pattern meaning due interpret spatial statist propos algorithm call slice st miner tackl algorithm design challeng spatial sequenc index preserv downward closur properti compar propos algorithm simpl algorithm call st miner util weak monoton properti sequenc index perform evalu synthet real world data set show slice st miner order magnitud faster st miner larg data set index term spatio tempor sequenti pattern densiti ratio sequenc index slice perform fast mine spatial colloc discov partial period pattern discret data sequenc abstract problem partial period pattern mine discret data sequenc find subsequ period frequent data sequenc essenti subproblem effici mine frequent pattern automat discoveri period correspond pattern previou method problem event sequenc databas assum period advanc requir addit databas scan comput period defin candid pattern work propos structur abbrevi list tabl alt effici algorithm comput period pattern requir small number pass perform studi present demonstr effect effici method discov minim infrequ structur xml document web servic choreographi scenario interoper bioinformat applic effici scalabl algorithm cluster xml document structur abstract standard xml inform exchang languag net huge amount inform format xml document order analyz inform effici decompos xml document store relat tabl popular practic queri process expens case excess number join requir recov inform fragment data collect consist document structur dtd mine cluster document allevi fragment problem propos hierarch algorithm grace cluster xml document base structur inform data notion structur graph graph propos support computation effici distanc metric defin document set document simpl metric yield cluster algorithm effici effect compar approach base tree edit distanc experi real data show algorithm discov cluster easili identifi manual inspect index term data mine cluster xml semistructur data queri process harp practic project cluster algorithm abstract high dimension data cluster exist subspac hide tradit cluster method number algorithm propos identifi project cluster reli user paramet guid cluster process cluster accuraci degrad incorrect valu real situat rare user suppli paramet valu accur practic difficulti appli algorithm real data paper analyz major challeng project cluster suggest algorithm depend heavili user paramet base analysi propos algorithm exploit cluster statu adjust intern threshold dynam assist user paramet result extens experi real synthet data method excel accuraci usabl outperform algorithm correct paramet valu artifici suppli encourag result suggest project cluster practic tool kind real applic index term data mine mine method algorithm cluster bioinformat optim data cube system design abstract design olap system support real time queri major research issu approach data cube materi precomput multidimension view data data warehous deriv set data cube answer frequent ask queri directli practic problem mainten cost data cube queri cost answer queri maintain data cube requir disk storag cpu comput mainten cost relat total size total number data cube materi case materi data cube impract mainten cost reduc merg data cube result larger data cube increas queri cost answer queri bound mainten cost queri cost strict user decid queri sacrif consider defin optim problem data cube system design mainten cost bound queri cost bound set frequent ask queri determin set data cube system answer largest subset queri violat bound np hard problem propos approxim greedi algorithm gr gm gmm shown effect effici experi censu data set forest cover type data set identifi project cluster gene express profil microarrai gene express data cluster hide subspac set regul gene similar express pattern subset sampl regul factor present express pattern dissimilar measur full input space tradit cluster algorithm make similar measur fail identifi cluster recent year number algorithm propos identifi kind project cluster reli critic paramet proper valu hard user determin paper algorithm dynam adjust intern threshold propos low depend user paramet allow user input domain knowledg experiment result show algorithm capabl identifi interest project cluster highli dusabl project cluster algorithm gene express profil project cluster hot research topic due abil cluster high dimension data exist project cluster algorithm depend critic user paramet determin relev attribut cluster case wrong paramet valu cluster perform degrad correct paramet valu rare real dataset paper propos project cluster algorithm depend user input determin relev attribut respond cluster statu adjust intern threshold dynam experiment result algorithm show higher usabl project cluster algorithm comparison studi work gene express dataset studi lymphoma high usabl algorithm encourag result suggest project cluster practic tool analyz gene express profil mainten partial dsum dbase histogram paper introduc effici method mainten wavelet base histogram built partial sum wavelet base histogram construct raw data distribut partial sum construct method merit previou work focus mainten raw data base histogram highli ineffici appli directli techniqu partial sum base histogram singl data updat trigger multipl partial sum turn trigger larg amount comput wavelet decomposit approxim string match dna sequenc approxim string match larg dna sequenc data import bioinformat studi shown suffix tree effici data structur approxim string match perform suffix arrai data structur store memori studi find suffix arrai suffix tree index dna sequenc data structur creat store disk due size propos auxiliari data structur greatli improv effici suffix arrai approxim string match problem extern memori model problem tackl parallel approxim match dna sequenc propos parallel algorithm problem implement pc cluster result show error allow small direct partit arrai machin cluster effici approach hand error allow larg partit data machin approach similar search set categor data signatur tree data mine applic analyz larg collect set data high dimension categor data search data type restrict classic problem mine associ rule classif similar search frequent appli oper access method multidimension numer data inappropri problem special index need propos method repres set data bitmap signatur organ hierarch index suitabl similar search relat queri type contrast previou techniqu signatur tree dynam reli hardwir constant experi synthet real dataset show robust data characterist scalabl databas size effici queri evalu iceberg distanc join filter index complex queri semi dstructur data web servic orient data process integr data process integr heterogen data sourc requir intens human resourc web servic standard programmat interfac applic applic commun gain consider momentum facilit process integr heterogen data sourc full advantag web servic propos web servic orient approach oper data sourc expos web servic base approach design declar script languag call wsipl web servic integr process languag special design languag construct integr process web servic systemat minim program effort final propos refer architectur applic exampl wsipl implement demonstr effect work effici algorithm increment updat frequent sequenc agraw srikant put forward problem mine frequent occur se quenc custom databas model custom databas consist set sequenc sequenc chronolog order set transact itemset effici algorithm increment updat concept space vocabulari problem inform retriev aris author index term con cept thesauru defin map relat term enhanc effect outlier detect low densiti pattern direct system mine data convers rule dispar data sourc quot explicit permiss provid full credit includ notic sourc paper download charg parallel mine outlier larg databas data mine import fast grow databas applic outlier except detect kind data mine appli varieti area monitor credit card fraud crimin activ electron commerc increas size attribut dimens databas previous propos detect method dimens longer applic automat construct onlin catalog topolog lattic dbase approach fo effici associ rule mine knowledg discoveri data mine pakdd mine confid rule support requir open problem find rule satisfi minimum confid necessarili minimum support support requir classic support base prune strat egi inapplic problem demand confidencebas prune strategi monoton confid call univers existenti upzuard closur hold rule size confid minimum confid attribut rule special size attribut confid support base prune bottleneck memori small store candid requir search implement strategi disk studi perform empir studi visual cluster valid method fastmap automat construct onlin catalog topolog ff fo deffici algorithm mine frequent sequenc index attribut reorder profil xml document filter inform deliveri cube model cluster analysi web access session adapt algorithm mine associ rule share dmemori parallel machin mine associ rule larg databas costli propos develop parallel algorithm task share memori multiprocessor smp propos parallel algorithm paradigm follow convent level wise approach iter length maximum larg itemset make matter wors impos synchron iter content share memori parallel system adapt asynchron parallel mine algorithm apm propos smp processor gener candid dynam count itemset support independ synchron optim techniqu propos reduct databas scan number candid algorithm apm implement sun enterpris share memori multiprocessor node experi show optim good effect apm substanti lead perfo pervas multimedia markup languag pmml xml dbase multimedia content specif pervas access build dens dregion dbase olap system line analyt process olap tool decis support system built data warehous rolap relat olap molap multidimension olap popular approach build olap system approach perform characterist molap good queri perform bad space effici rolap built matur rdbm technolog sizabl indic support data warehous small cluster multidimension data dens region spars point scatter rest space databas propos dens region locat separ spars point dens region subsequ repres small molap spars point put rolap tabl molap rolap approach integr structur build high perform space effici dens region base data cube pa adapt algorithm mine associ rule share dmemori parallel machin mine associ rule larg databas costli propos develop parallel algorithm task share memori multiprocessor smp propos parallel algorithm paradigm follow convent level wise approach iter length maximum larg itemset make matter wors impos synchron iter content share memori parallel system adapt asynchron parallel mine algorithm apm propos smp processor gener candid dynam count itemset support independ synchron optim techniqu propos reduct databas scan number candid algorithm apm implement sun enterpris share memori multiprocessor node experi show optim good effect apm substanti lead perfo discov reconcil conflict numer data integr built inform technolog capit fuel internet cost ectiv telecommun technolog led prolifer inform system dire exchang inform incap due lack semant interoper evid physic connect abil exchang bit byte longer adequ integr data autonom heterogen system call prior identif resolut semant conflict present requir system integr sift data dispar system painstak manner suggest process partial autom present methodolog techniqu discoveri potenti semant conflict underli data transform need resolv conflict methodolog begin classifi data conflict categori context independ context depend context independ conflict caus unexpect error context depend conflict primarili result heterogen underli data sourc facilit data integr data convers rule propos describ quantit relationship data valu involv context depend conflict gener approach propos discov data convers rule data approach consist major step relev attribut analysi candid model select convers function gener convers function select convers rule format implement prototyp system direct busi data statist base techniqu preliminari stu exploit dualiti maxim frequent itemset minim infrequ itemset fo effici associ rule mine interact approach build classif model cluster cluster valid densiti dbase mine quantit associ rule text dsourc discoveri gloss updat dynam web optim problem data cube system design olap system data cube precomput multidimension view data support real time queri reduc mainten cost relat number cube materi cube merg result larger cube increas respons time answer queri order satisfi mainten bound respons time bound user sacrific queri consider optim problem data cube system design optim initi set cube system answer maximum number queri satisfi bound np complet problem approxim algorithm greedi remov greedi merg propos experi censu databas result show approach effect effici introduct dss olap advanc data wareh technolog distribut scalabl xml document process architectur dcommerc system semant similar approach electron document model integr contextu semant autom web document search analysi anchor point index web document retriev tradit www search engin alta vista index recommend individu web page assist user locat relev document web grow number match page increas tremend rate user overwhelm larg answer set recommend search engin match document hypertext document structur destroi individu page compos document return logic start point hyper document hidden larg basket match page user spend lot effort brows page locat start point time consum process paper studi anchor point index problem set anchor point user queri small set kei page larger set document relev queri easili reach anchor point solv problem huge answer set effici rule dbase attribut dorient induct data mine data mine import techniqu tremend potenti commerci industri applic attribut orient induct power mine techniqu successfulli implement data mine system dbminer induct capabl limit uncondit concept gener paper extend concept gener rule base concept hierarchi enhanc greatli induct power previous propos induct algorithm appli gener rule base case problem induct anomali occur impact effici develop effici algorithm facilit induct rule base case avoid anomali perform studi shown algorithm superior previous propos algorithm base backtrack keyword data mine knowledg discoveri databas rule base concept gener rule base concept hier requir dbase data cube schema design line analyt process olap requir effici process complex decis support queri larg databas accept pre comput data cube reduc respons time queri dramat import design issu effici olap system choic data cube materi call problem data cube schema design problem paper show problem find optim data cube schema olap system limit memori np hard computation effici altern propos greedi approxim algorithm cmp variant algorithm cmp consist phase phase initi schema consist cube requir effici answer user queri form phase cube initi schema select merg satisfi memori constraint show cmp effect prun search space drolap dens dregion base approach dline analyt process rolap relat olap molap multidimension olap oppos techniqu build line analyt process olap system molap good queri perform suffer data distribut multidimension data cube spars rolap built matur rdbm technolog perform competit data warehous spars cluster multidimension data propos dens region base olap drolap system surpass rolap molap space effici queri perform drolap appli molap approach dens region discov data handl remain small percentag spars point rolap approach core build drolap system li mine dens region data cube defin dens region mine problem optim problem show convent cluster techniqu suitabl problem develop effici fast algorithm densiti dbase cluster larg databas lgen lattic dbase candid set gener algorithm fo effici associ rule mine algorithm associ rule mine variant basic apriori algorithm characterist apriori base algorithm candid itemset gener round size itemset increment round number databas scan requir apriori base algorithm depend size largest larg itemset paper devis gener candid set gener algorithm lgen gener candid itemset multipl size databas scan show reason set suggest larg itemset lgen significantli reduc number pass requir case pass suffici discov larg itemset irrespect size largest keyword data mine associ rule lattic apriori lgen introduct data mine recent attract consider attent databas practition research beca effect data distribut parallel mine associ associ rule mine import problem data mine crucial applic decis support market strategi propos effici parallel algorithm mine associ rule distribut share parallel system effici attribut incorpor power candid set prune techniqu techniqu distribut global prune sensit data distribut characterist data skew workload balanc prune effect skew balanc high implement fpm ibm sp parallel system perform studi show fpm outperform cd consist parallel version repres apriori algorithm result valid observ effect prune techniqu respect data distribut characterist show fpm nice scalabl par oblivi rout lc permut hypercub recommend anchor point structur dpreserv hypertext document retriev effect data skew parallel mine associ rule effici parallel algorithm fpm fast parallel mine mine associ rule share parallel system propos adopt count distribut approach incorpor power candid prune techniqu distribut prune global prune simpl commun scheme perform round messag exchang iter found prune techniqu sensit data skew describ degre uniform itemset distribut databas partit distribut prune effect data skew high global prune effect distribut prune mild data skew case implement algorithm ibm sp parallel machin perform studi confirm observ relationship effect prune techniqu data skew shown fpm outp asynchron parallel algorithm mine associ rule share dmemori multipprocessor sampl data mine case mainten discov associ rule natur sampl appeal techniqu data mine approxim solut case great satisfact user attempt sampl techniqu address problem maintain discov associ rule studi problem maintain discov associ rule updat made databas propos method examin chang part unchang part origin databas larg time wors updat rule perform frequent databas underli rule set chang effort wast paper devis algorithm emploi sampl techniqu estim differ associ rule databas databas updat estim differ determin updat discov user access pattern world wide web paper focu techniqu design level web tool system design architectur propos studi problem construct user profil user access histori gener increment techniqu maintain discov associ rule gener increment updat techniqu develop maintain associ rule discov databas case includ insert delet modif transact databas previous propos algorithm fup handl mainten problem case insert propos algorithm fup make previou mine result cut cost find rule updat databas insert case fup equival fup delet case fup complementari algorithm fup effici delet transact small part databas applic case gener case fup effici updat discov rule transact ad transact databas obsolet transact remov propos algorithm implement perform studi compar algorithm mine mainten discov associ rule updat paper devis algorithm estim differ associ rule databas updat estim differ determin updat mine associ rule estim differ larg time updat mine associ rule order discov learn rule discard estim differ small rule origin databas good approxim updat databas spend resourc updat rule accumul updat updat rule avoid overhead updat rule frequent introduct data mine attract attent practition research recent year combin techniqu field machin learn statist databas data mine enabl find usef discov reconcil semant conflict data mine perspect current approach semant interoper requir human intervent detect potenti con ict defin con ict resolv major imped achiev logic connect number dispar sourc larg paper demonstr detect reconcili semant conflict autom tool techniqu develop data mine commun describ process discov rule repres relationship semanticali relat attribut illustr effect approach exampl keyword databas integr data mine regress analysi semant conflict introduct avarieti onlin inform sourc receiv user applic emerg unpreced rate year contribut larg part exponenti growth internet advanc telecommun technolog intellig agent match inform provid consum world dwide dweb increment updat discov multi dlevel associ rule mainten discov associ rule larg databas increment updat techniqu increment updat techniqu develop mainten associ rule discov databas mine studi effici discoveri associ rule larg databas nontrivi maintain discov rule larg databas databas frequent occasion updat updat invalid exist strong associ rule turn weak rule strong studi increment updat techniqu propos effici mainten di cover associ rule transact data ad transact databas mainten discov knowledg case multi dlevel associ rule knowledg discoveri kdd system spend substanti amount effort search rule pattern larg amount data natur evolut consequ updat appli databas system updat previous discov knowledg reflect current state databas straight forward approach run discoveri process updat databas discov rule pattern cost effect gener unaccept case studi problem updat discov associ rule found nontrivi updat invalid exist strong associ rule turn weak rule strong increment techniqu fast algorithm fup propos previous updat discov singl level associ rule studi effici algorithm fup gener smaller fast distribut algorithm mine associ rule exist larg transact databas huge amount data high scalabl distribut system easi partit distribut central databas import investig effici method distribut mine associ rule studi disclos interest relationship local larg global larg itemset propos interest distribut associ rule mine algorithm fdm fast distribut mine associ rule gener small number candid set substanti reduc number messag pass mine associ rule perform studi show fdm superior perform direct applic typic sequenti algorithm perform enhanc lead variat algorithm effici mine associ rule distribut databas sequenti algorithm propos mine associ rule work mine associ rule distribut databas direct applic sequenti algorithm distribut databas effect requir larg amount commun overhead studi effici algorithm dma propos gener small number candid set requir messag support count exchang candid set number site distribut databas algorithm implement experiment test bed perform studi result show dma superior perform compar direct applic popular sequenti algorithm distribut databas mainten discov knowledg strategi updat associ rule abstract case dbase reason approach associ queri answer tradit dbm retriev data perfectli match user queri requir user detail databas schema desir obtain addit relev inform queri paper present method provid inform user explicitli ask domain specif knowledg queri depend user goal knowledg propos case base reason paradigm integr past user experi current goal order guid associ associ increment acquir observ past experi adapt answer current user queri prototyp associ queri answer system propos method implement top cooper databas system cobas ucla preliminari experiment result reveal feasibl scalabl method associ control introduct transact replic scheme replic databas node autonomi propos protocol replic databas central control transact transact site monitor remot data access transact commit approach broadcast transact remot site handl transact complet form site data type share privat data public data show transact work share privat data execut local concurr control protocol assum synchron network possibl partit failur show scheme transact execut manag commun delai compar central transact control motiv replic distribut databas protocol manag execut transact site access local remot data copi oper organ commit abort transact site refer site optim termin protocol distribut databas network partit site optim termin protocol distribut databas network partit 