approxim string search spatial databas abstract work present index structur mhrtree effici answer approxim string match queri larg spatial databas mhr tree base tree augment min wise signatur linear hash techniqu min wise signatur index node concis represent union gram string tree analyz prune function signatur base set resembl queri string gram tree index node mhr tree support wide rang queri predic effici includ rang nearest neighbor queri discuss estim rang queri select accur present adapt algorithm find balanc partit spatial string inform store tree extens experi larg real data set demonstr effici effect approach bed dtree dpurpos index structur string similar search base edit distanc weight set dbase string similar univers token weight databas consist string repres subset token queri string repres set token weight string similar queri identifi string databas similar queri larger user threshold weight string similar queri applic data clean integr find approxim match presenc typograph mistak multipl format convent data transform error show problem semant properti exploit design index structur support effici algorithm queri answer multi dcolumn foreign kei discoveri foreign primari kei relationship relat tabl import constraint databas data analysi perspect discov foreign kei crucial step understand work data foreign kei constraint data reason associ design inher data invalid due data inconsist work propos robust algorithm discov singl column multi column foreign kei previou work concentr discov singl column foreign kei varieti rule inclus depend column name minimum maximum valu propos gener rule term random subsum varieti rule develop effici approxim algorithm evalu random pass data final valid approach extens experi real synthet dataset method find frequent item data stream type dbase categor relat attribut work concentr categor relat attribut base data type assum attribut type characterist unknown unidentifi analyz compar varieti type base signatur classifi attribut base semant type data contain router identifi social secur number email address signatur subsequ applic cluster index optim compress applic case larg data collect gener distribut ungovern fashion end unknown incomplet inconsist complex schemata schema level meta data concentr heurist gener type base attribut signatur base local global comput approach show experiment decompos data gram signatur base gram distribut achiev good classif accuraci assumpt larg sampl data build signatur turn attent case small sampl data accur captur gram distribut data type imposs propos techniqu base dimension reduct soft cluster exploit correl attribut improv classif accuraci increment mainten length normal index approxim string match approxim string match problem receiv lot attent recent exist work inform retriev concentr varieti similar measur tf idf bm hmm specif tailor document retriev purpos applic depend retriev short string popular local search engin yellowpag yahoo local googl map index method need tailor short string purpos number index techniqu relat algorithm propos base length normal similar measur common denomin index length normal measur maintain underli structur presenc increment updat ineffici due data depend precomput weight distinct token string incorpor updat accomplish rebuild index regular time interv paper present framework advoc lazi updat propag kei featur effici increment updat immedi reflect data index strict guarante qualiti subsequ queri answer specif techniqu guarante fals neg limit number fals posit produc implement fulli work prototyp illustr propos idea work practic real dataset robust approxim aggreg sensor data manag system emerg area sensor base system signific challeng develop scalabl fault toler method extract inform data sensor collect approach data manag problem sensor databas system user perform aggreg queri min count avg read sensor network addit advanc queri frequenc count quantil estim support due energi limit sensor base network central data collect gener impract system network aggreg reduc network traffic aggreg strategi remain bandwidth intens combin fault toler multi path rout method environ avoid expens investig approxim network aggreg small sketch present duplic insensit sketch techniqu implement effici small sensor devic limit hardwar support analyz perform accuraci final present experiment evalu valid effect method find frequent item stream data doi frequent item problem process stream item find occur fraction time heavili studi problem mine data stream date back applic reli directli indirectli find frequent item implement larg scale industri system paper describ import algorithm problem common framework place solut histor context describ connect aim clarifi confus surround properti illustr properti algorithm provid baselin implement give empir evid consider variat perform frequent item algorithm method implement find frequent item high accuraci ten kilobyt memori rate million item cheap modern hardwar letter special issu editor publish quarterli distribut tc member scope includ design implement model theori applic databas system technolog letter confer inform new effici approxim search string collect tutori comprehens overview recent research progress import problem approxim search string collect identifi exist index search algorithm filter strategi select estim techniqu work comment respect merit limit repeat workabl evalu sigmod fast index algorithm set similar select queri abstract data collect inconsist aris due varieti reason desir identifi resolv effici set similar queri commonli data clean match similar data work concentr set similar select queri queri set retriev set collect similar greater threshold set similar measur propos past data clean purpos work concentr weight similar function tf idf introduc variant suit set similar select relat databas context variant special semant properti exploit design effici index structur algorithm answer queri effici present modif exist technolog work set similar select queri introduc algorithm base threshold algorithm exploit semant properti similar measur achiev perform theori practic random synops queri assur data stream due overwhelm flow inform data stream applic compani acquir resourc deploi data stream manag system dsm choos altern outsourc data stream desir comput parti data outsourc remot comput intrins rais issu trust make outsourc queri assur data stream problem import practic implic set continu group sum queri process remot untrust server client limit process capabl observ stream server regist queri server dsm receiv result request client verifi integr result significantli fewer resourc evalu queri local goal propos probabilist verif algorithm select aggreg group queri constant space irrespect result set size low updat cost stream element arbitrarili small probabl failur gener algorithm toler number erron group detect order support semant load shed server discuss hard support random load shed final implement techniqu perform empir evalu live network traffic trip plan queri road network databas trip plan queri road network databas index structur extens enabl effici similar search larg databas index techniqu linear transform scheme reduc dimens fast approxim reduct approach approxim unbound approxim volum extend dataspac estim retriev set impair perform paper present linear transform scheme extract import paramet data prove paramet correspond bound volum search sphere irrespect dimension special workspac map mechan deriv tight bound paramet prove result highlight insight problem propos solut formul measur lower bound euclidean distanc discuss implement techniqu popular index structur extens experi confirm superior techniqu recent state art scheme index spatio dtempor archiv spatio tempor object object evolv time applic due natur applic store evolut object time order answer histor queri queri refer past state evolut requir larg special databas term articl spatio tempor archiv effici process histor queri spatio tempor archiv requir equal sophist index scheme typic spatio tempor index techniqu repres object minimum bound region mbr extend tempor dimens index tradit multi dimension index structur rough mbr approxim introduc excess overlap index node deterior queri perform articl introduc robust index scheme answer spatio tempor queri effici number algorithm heurist elabor preprocess spatiotempor archiv order produc finer object approxim combin multi version index structur greatli improv queri perform comparison straightforward approach propos techniqu introduc queri effici space tradeoff tune structur resourc empir observ estim amount addit storag space requir improv queri perform factor provid heurist appli propos idea onlin set discuss final experiment evalu conduct show merit propos techniqu dtree dynam index structur spatial search hash sampl select estim set similar select queri studi select estim techniqu set similar queri wide varieti similar measur set propos past work concentr class weight similar measur tf idf bm cosin similar variant design select estim base priori construct sampl studi pitfal straightforward applic random sampl argu care sampl construct uniform random sampl yield low accuraci queri sensit realtim sampl expens exact solut cpu cost show build robust sampl priori base exist synops distinct estim prove accuraci techniqu theoret verifi perform experiment algorithm order magnitud faster exact solut small space overhead find frequent item data stream abstract present pass algorithm estim frequent item data stream limit storag space method reli data structur call count sketch estim frequenc item stream algorithm achiev space bound previou algorithm problem natur distribut item frequenc addit algorithm lead directli pass algorithm problem estim item largest absolut chang frequenc data stream knowledg problem previous studi literatur norm paper propos object measur base inform theoret notion mutual inform estim word associ norm comput readabl corpora standard method obtain word associ norm test thousand subject hundr word costli unreli propos measur associ ratio estim word associ norm directli comput readabl corpora make estim norm ten thousand word continu constraint queri evalu spatiotempor stream abstract paper studi evalu continu constraint queri ccq spatiotempor stream ccq trigger alert configur constraint stream event space time satisfi instanc server receiv updat gp enabl agent report posit measur environment read ccq alert read closer km time differ minut report high pressur low temperatur model ccq constraint satisfact problem csp develop solut continu evalu techniqu fast arriv rate incom event minim memori requir predefin window constraint util structur queri order show merit propos techniqu implement system prototyp evalu real data proof dinfus stream enabl authent slide window queri stream queri dsensit embed common problem type databas retriev similar match queri object find match larg databas slow practic domain object compar computation expens similar distanc measur paper propos method approxim nearest neighbor retriev space method embed base mean construct function map object real vector space map preserv larg amount proxim structur origin space rapidli obtain short list match queri main novelti method construct embed queri sensit distanc measur measur distanc vector space term querysensit mean distanc measur depend current queri object report experi imag databas handwritten digit time seri databas case propos method outperform exist state art embed method mean significantli trade cienci retriev accuraci random sampl continu stream arbitrari updat global distanc dbase segment trajectori work introduc distanc base criteria segment object trajectori segment lead simplif origin object smaller complex primit suit storag retriev purpos previou work trajectori segment attack problem local segment separ trajectori databas directli optim inter object separ mine oper search cluster classif larg databas paper analyz trajectori segment problem global perspect util data awar distanc base optim techniqu optim pairwis distanc estim lead effici object prune deriv exact solut distanc base formul due intract complex exact solut present approxim greedi solut exploit forward search local optim solut greedi solut impos prohibit comput cost put forward lightweight varianc base segment techniqu intellig relax pairwis distanc area affect mine oper index spatiotempor archiv dynam authent index structur outsourc databas abstract outsourc databas odb system databas owner publish data number remot server goal enabl client edg network access queri data effici server untrust compromis queri authent essenti compon odb system chapter present techniqu authent select rang queri analyz perform cost metric addit discuss extens queri type index multidimension time dseri index spatiotempor archiv time relax spatiotempor trajectori join effici trajectori join symbol represent rpj produc fast join result stream rate dbase optim conceptu partit effici method continu nearest neighbor monitor set object queri point nearest neighbor nn queri retriev object lie closest problem studi static dataset tradit method extend highli dynam environ multipl continu queri requir real time result object queri receiv frequent locat updat paper propos conceptu partit cpm comprehens techniqu effici monitor continu nn queri cpm achiev low run time handl locat updat object fall vicin queri ignor rest multipl static move queri make assumpt object move pattern analyz perform cpm show outperform current state art algorithm problem set final extend framework aggreg nn ann queri monitor data object minim aggreg distanc respect set queri point object minimum sum distanc queri point trip plan queri spatial databas paper discuss type queri spatial databas call trip plan queri tpq set point interest space point belong specif categori start point destin tpq retriev trip start pass point categori end driver travel boston provid stop ga station bank post offic goal provid rout term distanc traffic road condit difficulti queri li exist multipl choic categori paper studi fast approxim algorithm tpq metric space provid number approxim algorithm approxim ratio depend number categori maximum number point complex spatio dtempor pattern queri paper introduc type queri spatio tempor pattern queri stp queri specifi spatio tempor pattern sequenc distinct spatial predic predic tempor order exact rel matter stp queri type spatial predic rang search nearest neighbor predic exact tempor constraint time instant time interv gener rel order queri predic tradit spatio tempor index structur type queri ineffici applic solut altern propos special queri evalu algorithm stp queri time present index structur suitabl stp queri order final conduct comprehens experiment evalu show merit techniqu index multi ddimension trajectori similar queri sail spatial index librari effici applic integr 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 spatio dtempor data servic share dnoth environ sail librari effici applic integr spatial indic scientif applic deal spatial spatiotempor multidimension index structur typic manag million object arbitrari complex featur choos method index data difficult index librari combin indic program interfac valuabl paper present sail spatial index librari robust extens librari enabl simpl integr spatial index structur exist applic focu design issu elabor techniqu make framework gener support user defin data type customiz spatial queri broad rang spatial spatio tempor index structur librari publicli successfulli util research commerci applic perform evalu spatio dtempor select estim techniqu abstract spatio tempor applic deal move object environ databas typic maintain initi posit move function object updat databas object posit manag updat issu move function deviat threshold simplic assum object move linear trajectori maintain move function databas introduc problem databas answer queri object posit futur find object area minut paper present perform evalu techniqu estim select queri exist estim store main memori updat dynam propos approach techniqu histogram secondari index base estim run divers set experi identifi strength weak approach wide varieti dataset dline discoveri dens area spatio dtempor databas abstract move object databas receiv consider attent recent previou work concentr model index problem queri select estim introduc problem address densiti base queri spatio tempor domain find region object ten minut user interest find time period interv queri answer remain valid formal defin class densiti base queri give approxim line techniqu answer effici typic threshold region consid dens part queri difficulti problem li fact spatial tempor predic queri techniqu introduc find candid dens region time futur make scalabl subdivid spatial univers grid limit queri pre time horizon final valid approach experiment evalu effici index spatiotempor object spatiotempor object object chang posit extent time applic paper examin problem index larg volum data import environ spatiotempor object move chang gener case object movement defin combin polynomi function concentr snapshot small interv queri common examin histori gather data obviou approach approxim spatiotempor object mbr tradit multidimension access method index ineffici object live long time interv larg mbr introduc lot empti space cluster long interv dealt tempor databas partial persist indic differenti problem tradit tempor index object allow move chang lifetim wai need approxim gener spatiotempor object obviou solut introduc artifici split lifetim long live object split smaller consecut piec decreas empti space increas number index mbr give optim algorithm heurist split spatiotempor object predefin number piec upper bound total number split present algorithm decid split distribut object total empti space minim number split increas indefinit extra object eventu affect queri perform usi index anim object spatiotempor access method abstract present approach index anim object effici answer queri posit time space anim movi spatiotempor evolut movi view order sequenc frame frame space occupi object frame queri interest rang queri form find object area frame fi fj nearest neighbor queri find nearest object posit frame fi fj straightforward approach index object consid frame sequenc dimens access method tree variant assign long lifetim interv object consecut frame long interv difficult cluster effici index propos reduc problem partial persist problem access method made partial persist show approach lead faster queri perform storag proport total number frame evolut differenti problem tradit tempor index approach object allow move chang extent continu frame present method approxim object evolut formul optim problem provid optim solut case object move linearli final present extens experiment studi propos method concentr anim movi approach gener appli spatiotempor applic index term access method spatiotempor databas anim object multimedia 