compar structur rna multipl align proceed acm sigmod intern confer manag data bioinformat databas detect conserv secondari structur rna molecul constrain structur align journal homepag www elsevi locat compbiolchem discov frequent agreement subtre phylogenet data abstract studi data mine problem discoveri frequent agreement subtre fast set phylogenet tree phylogenet tree phylogeni unord tree order sibl unimport leaf tree label repres taxon speci organ intern node unlabel tree root repres common ancestor speci tree unroot unroot phylogeni aris due lack suffici evid infer common ancestor taxa tree fast problem address natur extens maximum agreement subtre mast problem wide studi comput phylogenet commun paper establish framework tackl fast problem root unroot phylogenet tree data mine techniqu develop canon form root tree phylogeni awar tree expans scheme gener candid subtre level level present effici algorithm find fast set root tree apriori approach show correct complet propos method final discuss extens techniqu unroot tree experiment result demonstr propos method work capabl find interest pattern synthet data real phylogenet tree index term data mine evolutionari bioinformat comput phylogenet algorithm design pattern discoveri analysi biolog data soft comput approach constrain rna structur align algorithm applic motif detect untransl region trypanosoma brucei mrna detect conserv rna secondari structur viral genom radar approach integr rna motif databas studi phylogenet tool genom nomenclatur data clean classif rna sequenc support vector machin biolog data clean case studi radar web server rna data analysi research kernel method rna classif radar interactiveweb dbase toolkit rna data analysi research mine frequent agreement subtre phylogenet databas kernel design rna classif support vector machin data mine bioinformat bioinformat studi applic comput method life scienc data present enjoi surg interest main reason public near complet sequenc human genom anticip knowledg deriv process great impact modern medicin pharmaceut industri expect util knowledg drug design interest bioinformat structur data biomonitor phylogenet anomali aggreg system lineag path integr phylogenet resourc introduct data mine bioinformat method align rna secondari structur applic rna motif detect metricmap embed techniqu process distanc dbase queri metric space abstract paper present embed techniqu call metricmap capabl estim distanc pseudometr space databas object distanc function object pseudometr map object vector pseudo euclidean space low dimens preserv distanc object approxim embed techniqu approxim oracl process broad class distanc base queri adapt data mine applic data cluster classif present theori underli metricmap conduct experi compar metricmap method includ mvp tree tree process distanc base queri experiment result protein rna data show good perform superior metricmap method index term bioinformat data mine embed method metric space nearest neighbor similar search xml cluster retriev princip compon analysi unord tree mine applic phylogeni frequent structur mine fsm aim discov extract pattern frequent occur structur data tree graph fsm find applic bioinformat xml process web log analysi paper present fsm techniqu find pattern root unord label tree pattern interest cousin pair tree cousin pair pair node share parent grandpar great grandpar tree algorithm find interest cousin pair flowmin find flow pattern spatio dtempor databas xml cluster princip compon analysi genescout data mine system predict vertebr gene genom dna sequenc bio dajax extens framework biolog data clean proceed acm sigkdd workshop data mine bioinformat biokdd comput biologi genom informat rna profil analysi techniqu proteom yield vast amount data gene express protein level point develop methodolog identifi analyz complex biolog network chapter describ develop java base tool help dynam find visual metabol network tool consist part part text mine tool pull potenti metabol relationship pubm databas relationship review domain expert ad exist network model result visual interact graph displai modul basic metabol regulatori flow network model fuzzi cognit map causal connect pull sequenc data genet algorithm base logic proposit gener search tempor pattern microarrai data exampl regulatori open remark treerank similar measur nearest neighbor search phylogenet databas phylogenet tree unord label tree leaf node label order sibl unimport paper propos similar measur call treerank phylogenet tree present algorithm comput treerank score queri tree pattern data tree treerank score measur topolog relationship found similar propos algorithm calcul treerank score time number node appear number node develop search engin queri pattern databas tree find rank nearest tree neighbor near measur propos similar function structur base search engin fulli oper world wide web protein classif geometr hash approach data mine bioinformat report biokdd proceed acm sigkdd workshop data mine bioinformat biokdd cluster classifi enzym metabol pathwai preliminari result foreword refer set intern guidelin recommend report present statist data metadata handbook prepar primarili short term econom statist recommend present relev annual structur statist social popul statist handbook draw heavili exist intern statist guidelin recommend develop twenti year intern organis nation statist agenc intern standard develop implement intern organis nation agenc develop develop economi articul comprehens set recommend report present statist data metadata aspect demand improv intern compar consist statist compil publish nation agenc intern organis requir driven closer econom social linkag oecd member countri countri larg emerg member economi discuss handbook major differ statist compil countri flow definit concept mine gene dna genescout algorithm applic tree graph search modern search engin answer keyword base queri extrem effici impress speed due clever invert index structur cach domain independ knowledg string thousand machin research effort attempt gener keyword search keytre keygraph search tree graph applic gener databas system paper survei algorithm applic give emphasi work structur dbase search engin phylogenet databas atreegrep approxim search unord tree unord label tree tree node string label parent child relationship signific order sibl unimport paper present approach find pattern ddimension graph algorithm applic scientif data mine paper present method find pattern dimension graph node graph undecompos atom unit label edg link atom unit pattern rigid substructur occur graph allow arbitrari number structur rotat translat small number user edit oper pattern graph pattern appear graph graph modifi call appear approxim occurr edit oper includ relabel node delet node insert node propos method base geometr hash techniqu hash node triplet graph tabl compress label triplet tabl demonstr util algorithm discuss applic scientif data mine xml queri find approxim pattern undirect acycl graph approxim pattern match problem undirect acycl graph specif pattern graph data graph integ present algorithm locat subgraph distanc distanc measur degre metric publish previous time complex algorithm amp radic log number node min maximum degre central algorithm procedur find approxim pattern root unord tree freeli allow cut discuss biokdd workshop data mine bioinformat biokdd recent advanc data minig bioinformat proceed acm sigkdd workshop data mine bioinformat biokdd mine world wide web inform search approach techniqu extract featur protein sequenc paper propos techniqu extract featur protein sequenc featur input bayesian neural network bnn appli bnn classifi protein sequenc obtain pir protein databas maintain nation biomed research foundat evalu perform propos approach compar protein classier built base sequenc align machin learn method experiment result show high precis propos classi er complementar bioinformat tool studi paper dna sequenc classif expect maxim algorithm neural network case studi paper present techniqu biosequ classif focu recogn coli promot dna specif unlabel dna sequenc determin coli promot expect maxim em algorithm locat bind site coli promot sequenc em algorithm differ previous publish em algorithm assum uniform distribut length spacer bind site bind site spacer bind site transcript start site algorithm deduc probabl distribut length base locat bind site select featur coli promot sequenc inform content repres featur orthogon encod method feed featur neural network promot recognit empir studi show propos approach achiev good perform dataset bioinformat introduct special issu effect hidden markov model detect splice junction site dna sequenc precis environment search integr hierarch inform search envirodaemon inform retriev evolv search refer abstract document search web involv search engin promis pars full text file audio video multimedia index web million page grow difficulti locat relev inform appar preval mean inform retriev reli syntax base method keyword string charact present search engin return match document method satisfactori easi implement inher limit make unsuit task syntact pattern user interest keyword mean locat word titl header paper describ precis search approach environment domain locat inform syntact criteria augment util inform context main emphasi paper li treatment structur knowledg essenti aspect topic interest encod individu item relationship exampl structur knowledg hypertext document diagram logic chemic formula benefit approach enhanc precis approxim search focus context specif search engin environ envirodaemon find similar consensu tree algorithm distanc hierarchi applic neural network biolog data mine case studi protein sequenc classif biolog data mine aim extract signific inform dna rna protein signific inform refer motif function site cluster classif rule paper present biolog data mine classif protein sequenc neural network propos techniqu extract featur protein data combin bayesian neural network classifi protein sequenc obtain pir protein databas maintain nation biomed research foundat evalu perform propos approach compar protein classifi built base sequenc align machin learn method experiment result show high precis propos classifi complementar tool studi paper keyword bioinformat biolog data mine machin learn neural network sequenc align featur extract protei approxim search engin structur databas identifi consensu tree align fast similar search ddimension structur databas paper studi similar search problem class relat queri index structur data mine cluster abstract paper present index structur call metric map take set object distanc metric map object dimension space distanc object approxim preserv index structur tool cluster visual data intens applic replac expens distanc calcul sum squar calcul make cluster larg databas expens distanc metric practic compar index structur data mine index structur fastmap recent propos faloutso lin criteria rel error cluster accuraci rel error show astmap lower rel error metricmap euclidean distanc ii metricmap alow rel error astmap euclidean distanc gener distanc metric iii combin reduc error similar result obtain compar accuraci pattern discoveri biomolecular data tool techniqu applic applic bayesian neural network protein sequenc classif paper present applic neural network biomed data mine specif propos hybrid approach combin similar search bayesian neural network classifi protein sequenc appli techniqu recogn globin sequenc obtain databas maintain protein inform resourc pir nation biomed research foundat experiment result excel perform propos approach keyword artifici intellig biomed applic data mine machin learn introduct data mine knowledg discoveri data kdd aim find signific inform set data knowledg mine dataset refer pattern associ rule belief network classif cluster regul techniqu mine knowledg borrow disciplin includ pattern recognit machin learn gene evalu class distanc dmap algorithm data mine cluster distanc map algorithm take set object distanc metric map object euclidean pseudo euclidean space distanc object approxim preserv distancemap algorithm tool cluster visual data intens applic replac expens distanc calcul sum squar calcul make cluster larg databas expens distanc metric practic paper present distanc map algorithm conduct experi compar perform data cluster applic includ algorithm call fastmap metricmap hybrid heurist combin algorithm wai experiment result synthet rna data show superior hybrid algorithm result impli fastmap metricmap captur complementari inform distanc metric ca framework biolog pattern discoveri network workstat pattern discoveri classif biosequ identifi approxim common substructur tree base restrict edit distanc biocomput introduct special issu techniqu dna sequenc classif dna sequenc classif activ determin unlabel sequenc belong exist class paper propos techniqu dna sequenc classif techniqu work compar unlabel sequenc group activ motif discov element distinct element techniqu gener match gap fingerprint element experiment result obtain run algorithm long conserv alu sequenc demonstr good perform present method compar fasta appli conserv short function site splice junction variat techniqu combin fingerprint consensu sequenc analysi result current classifi emploi text compress machin learn algorithm introduct dna sequenc classif import approxim oracl distanc metric space algorithm find largest approxim common substructur tree order label tree tree node label left order children fix tree applic vision pattern recognit molecular biologi natur languag process substructur order label tree connect subgraph order label tree integ largest approxim common substructur problem find substructur substructur edit distanc exist substructur satisfi distanc constraint sum size greater sum size present dynam program algorithm solv problem run fast fastest algorithm comput edit distanc tree distanc allow common substruc scientif data mine case studi graphic environ chang detect structur document electron enterpris engin outlin architectur paper put vision organiza tion fulli embmc comput support propos busi process orient architectur electron enterpris engin eee enabl enter prise manag evolv technolog orga nizat process efect integr manag enterpris inform electron empow knowledg worker level broad decis port capabl goal eee architectur empow enterpris make format asset oper efect tha era electron commerc part project develop standard base customiz integr tool set call support environ enterpris engin seee paper present current seee architectur show support eee goal seee architectur eee drive electron commerc push compani move opera tion line organ deal client custom line handl telecommut work group distribut moss countri world wide deal govern agenc organ relat line compani move mission critic portion oper line fast retriev electron messag mistyp word spell error abstract paper present index structur retriev electron messag mistyp word spell error queri string search kei find messag approxim queri insert delet mismatch allow match queri word phrase messag approach store messag sequenti databas hash fingerprint number fingerprint file queri fingerprint hash file histogram vote construct messag deriv lower bound base prune larg number nonqualifi messag vote lower bound search paper present experiment result demonstr effect index structur lower bound scientif data classif case studi autom discoveri activ motif dimension molecul paper present method discov approxim common motif activ motif dimension molecul node molecul repres point euclidean space edg repres undirect line segment connect node molecul motif rigid substructur occur molecul allow arbitrari number rotat translat small number user node insert delet oper motif molecul call approxim occurr propos method combin geometr hash techniqu block detect algorithm undirect graph demonstr util algorithm discuss applic classifi famili molecul pertain antibacteri sulfa drug anti anxieti agent benzodiazepin antiadrenerg agent fi receptor experiment result structur match discoveri document databas memo written sgml markup minim tag visibl carbon copi recipi sgml lt doctyp memo system gt fig sgml memo document charl goldfarb fast retriev electron messag mistyp word spell error abstract paper present index structur retriev electron messag mistyp word spell error queri string search kei find messag approxim queri insert delet mismatch allow match queri word phrase messag approach store messag sequenti databas hash fingerprint number fingerprint file queri fingerprint hash file histogram vote construct messag deriv lower bound base prune larg number nonqualifi messag vote lower bound search paper present experiment result demonstr effect index structur lower bound knowledg discov document classif tree match texpro autom discoveri activ motif multipl rna secondari structur scientif data mine case studi visual tool pattern match discoveri scientif databas sequenc databas compris sequenc data linear structur descript natur entiti approxim pattern discoveri sequenc databas lead import conclus predict phenomena tradit databas technolog suitabl accomplish task techniqu develop dissert propos techniqu discov pattern sequenc databas techniqu incorpor pattern match algorithm heurist discoveri optim experiment result appli techniqu gener data dna protein show effect propos techniqu develop classifi pattern discoveri algorithm previous publish fingerprint techniqu appli classifi classifi dna protein seq curriculum knowledg represent manipul knowledg dbase tutor system abstract knowledg base tutor system kbt comput base instruct system artifici intellig techniqu peopl learn subject found knowledg commun process involv kbt human student decompos seri commun cycl cycl concentr topic major phase plan discuss evalu remedi major contribut work develop gener architectur support knowledg commun kbt student graphic notat schema support curriculum knowledg represent manipul plan phase tutor process curriculum knowledg tutor system determin sequenc topic discuss student effect diagnos student mistak curriculclm knowledg base goal structur prerequisit relat multipl wai organ topic focus develop sql tutor kbt domain sql program system featur effici control mechan explicit curriculum knowledg represent individu privat tutor allow student rel freedom decid studi domain knowledg subject system student group oper hand tailor learn schedul special background request interest index term omput educ intellig tutor knowledg represent manipul topic associ graph learn graph curriculum manag inform extract structur part offic document edit distanc undirect acycl graph problem compar cual graph connect undirect acycl graph node label problem motiv studi inform retriev bio chemic molecular databas suppos defin distanc cual graph weight number edit oper insert node delet node relabel node transform reduct exact cover set show find distanc cual graph np complet view hard problem propos constrain distanc metric call degre distanc requir node insert delet neighbor metric present effici algorithm solv problem algorithm run time amp gener weight edit oper time amp radic log integr weight edit oper number node min maximum degre lhm logic dbase hypertext data model integr hypertext inform system edit distanc undirect acycl graph relat problem lhm logic dbase hypertext data model integr hypertext inform system integr toolkit pattern match pattern discoveri scientif sequenc databas compris sequenc data linear structur descript natur entiti approxim pattern discoveri sequenc databas lead import conclus predict phenomena tradit databas technolog suitabl accomplish task techniqu develop dissert propos techniqu discov pattern sequenc databas techniqu incorpor pattern match algorithm heurist discoveri optim experiment result appli techniqu gener data dna protein show effect propos techniqu develop classifi pattern discoveri algorithm previous publish fingerprint techniqu appli classifi classifi dna protein seq pattern match pattern discoveri scientif sequenc databas compris sequenc data linear structur descript natur entiti approxim pattern discoveri sequenc databas lead import conclus predict phenomena tradit databas technolog suitabl accomplish task techniqu develop dissert propos techniqu discov pattern sequenc databas techniqu incorpor pattern match algorithm heurist discoveri optim experiment result appli techniqu gener data dna protein show effect propos techniqu develop classifi pattern discoveri algorithm previous publish fingerprint techniqu appli classifi classifi dna protein seq approach model person offic document algorithm approxim graph match sequenc databas compris sequenc data linear structur descript natur entiti approxim pattern discoveri sequenc databas lead import conclus predict phenomena tradit databas technolog suitabl accomplish task techniqu develop dissert propos techniqu discov pattern sequenc databas techniqu incorpor pattern match algorithm heurist discoveri optim experiment result appli techniqu gener data dna protein show effect propos techniqu develop classifi pattern discoveri algorithm previous publish fingerprint techniqu appli classifi classifi dna protein seq document manag hypertext logic model approach work emploi logic model common languag integr independ inform domain document manag hypertext concept hypertext concern inform manag ment access hypertext consid knowledg represent model structur inform system serv hypertext en gine give user flexibl access infor mation command underli applic tailor applicatiov structur paper fo cuse integr hypertext function intellig person document process system call texpro develop independ logic data mod el hypertext texpro element present logic represent bridg law map found logic model invalu tool understand domain eflect integr knowledg dbase tutor system sql program combinatori pattern discoveri scientif data preliminari result sequenc databas compris sequenc data linear structur descript natur entiti approxim pattern discoveri sequenc databas lead import conclus predict phenomena tradit databas technolog suitabl accomplish task techniqu develop dissert propos techniqu discov pattern sequenc databas techniqu incorpor pattern match algorithm heurist discoveri optim experiment result appli techniqu gener data dna protein show effect propos techniqu develop classifi pattern discoveri algorithm previous publish fingerprint techniqu appli classifi classifi dna protein seq approxim graph match probabilist hill climb algorithm sequenc databas compris sequenc data linear structur descript natur entiti approxim pattern discoveri sequenc databas lead import conclus predict phenomena tradit databas technolog suitabl accomplish task techniqu develop dissert propos techniqu discov pattern sequenc databas techniqu incorpor pattern match algorithm heurist discoveri optim experiment result appli techniqu gener data dna protein show effect propos techniqu develop classifi pattern discoveri algorithm previous publish fingerprint techniqu appli classifi classifi dna protein seq docflow event ddriven visual program environ offic autom document process combinatori pattern discoveri scientif data preliminari result sequenc databas compris sequenc data linear structur descript natur entiti approxim pattern discoveri sequenc databas lead import conclus predict phenomena tradit databas technolog suitabl accomplish task techniqu develop dissert propos techniqu discov pattern sequenc databas techniqu incorpor pattern match algorithm heurist discoveri optim experiment result appli techniqu gener data dna protein show effect propos techniqu develop classifi pattern discoveri algorithm previous publish fingerprint techniqu appli classifi classifi dna protein seq system approxim tree match order label tree tree node label left order children fix tree applic vision pattern recognit molecular biologi program compil natur languag process applic involv compar tree retriev extract inform repositori tree exampl includ classif unknown pattern analysi newli sequenc rna structur semant taxonomi dictionari definit gener interpret nonprocedur program languag automat error recoveri correct program languag previou system exact match gener regular express match tree comparison paper present system call approxim tree atb inexact match tree atb system interact user simpl power queri languag graphic devic exact approxim algorithm unord tree match approxim tree match presenc variabl length don care order label tree tree sibl order matter paper present algorithm problem approxim match tree variabl length don care vldc string vldc symbol pattern substitut symbol data string comer pattern substitut substr put match data string comput approxim vldc match string mean substitut pattern data string match allow comer match counter distanc repres cost remov comer substitut unt gener approxim vldc string match algorithm approxim vldc match tree time complex algorithm jp theta jdj theta min depth leav theta min de formal approach model offic inform system tool classifi offic document oi ce document retriev system capabl process incomplet vagu queri intellig document support environ research issu uncertain queri process offic document retriev system fast serial parallel algorithm approxim tree match vldc order label tree tree node label left order children fix tree applic vision pattern recognit molecular biologi program compil natur languag process applic involv compar tree retriev extract inform repositori tree exampl includ classif unknown pattern analysi newli sequenc rna structur semant taxonomi dictionari definit gener interpret nonprocedur program languag automat error recoveri correct program languag previou system exact match gener regular express match tree comparison paper present system call approxim tree atb inexact match tree atb system interact user simpl power queri languag graphic devic pattern match unord tree texpro intellig document process system optim equijoin queri distribut databas relat hash partit class distribut databas system consist set node connect high bandwidth network node consist processor random access memori slower larger memori disk share memori node data horizont partit hash function descript character parallel distribut databas system recent propos commerci academ studi optim problem aris queri processor repartit relat intermedi result particip multijoin queri estim size intermedi relat show optimum solut close chain queri np complet optim problem star tree gener graph queri effect heurist hard case gener approach result extend attribut partit scheme sort partit attribut partit object databas queri process distanc metric applic vision molecular biologi common problem find similar object target distanc measur larg databas paper present scheme queri pro cess situat basic strategi par tialli precomput inter object distanc distanc inform triangl inequ elimin calcul object distanc evalu queri propos heurist speed queri evalu seri experi ment perform evalu effect scheme rel perform heurist queri final investig possibil iti parallel scheme simul result show parallel appli stage evalu queri techniqu dmatch retriev scheme answer match queri file collect object match queri find object file closest di similar measur target previou work suggest reduc number comparison requir achiev desir result triangl inequ start data structur file reflect precomput intrafil distanc gener techniqu optimum set precomput intrafil distanc empir result present illustr effect scheme perform rel previou algorithm 