assess expertis awar resolut network content dawar resolut sequenc mine ticket rout mine divers network top dk aggreg queri larg network synthes doptim malwar specif suspici behavior gener model ticket resolut expert network frequent trajectori mine gp data cross dsell optim custom promot proxim pattern mine larg graph discrimin frequent subgraph mine optim guarante scalabl olap mine inform network smallblu social network analysi expertis search collect intellig identifi bug signatur discrimin graph mine doptim supervis featur select frequent subgraph graph classif increasingli import step numer applic domain function predict molecul protein computeris scene analysi anomali detect program flow approach propos literatur graph classif base frequent subgraph popular branch graph repres binari vector compon indic graph subgraph frequent dataset larg graph face enorm problem number frequent subgraph grow exponenti size graph possess discrimin power make mine chemic data increas demand storag scientif data demand effici retriev mechan analyt report purpos case design databas support scientif data rightli driven scienc scienc centric view data manag make subsequ report inform easi natur solut problem data warehous approach scientif data manag graph olap multi ddimension framework graph data analysi effect present graph pattern structur repres approach research interest gener interest data mine data wareh databas system current research focus model manag analyz larg scale graph inform network data direct discrimin pattern mine effect classif classif graph olap onlin analyt process graph research interest gener interest data mine data wareh databas system current research focus model manag analyz larg scale graph inform network data effici ticket rout resolut sequenc mine mine signific graph pattern leap search easyticket ticket rout recommend engin enterpris problem resolut manag problem ticket kei issu servic industri larg servic provid handl thousand problem ticket custom daili basi effici process ticket highli depend ticket rout transfer problem ticket expert group search resolv ticket ticket manag system ticket rout system manual oper support personnel demo introduc easyticket ticket rout recommend engin help autom process mine ticket histori data model enterpris social network repres function relationship expert group ticket rout base network system rout recommend ticket experiment studi million real world problem ticket show averag easyticket improv effici ticket rout mine coloss frequent pattern core pattern fusion extens research frequent pattern mine past decad brought number pattern mine algorithm effect effici exist frequent pattern mine algorithm encount challeng mine larg pattern call coloss frequent pattern presenc explos number frequent pattern coloss pattern critic applic domain bioinformat studi investig mine approach call pattern fusion effici find good approxim coloss pattern pattern fusion coloss pattern discov fuse small core pattern step increment pattern growth mine strategi adopt apriori fp growth examin larg number mid size properti distinguish pattern fusion exist frequent pattern mine approach draw mine methodolog empir studi show case current mine algorithm proce pattern fusion mine result set close approxim complet set coloss pattern qualiti evalu model propos paper discrimin frequent pattern analysi effect classif applic frequent pattern classif appear sporad studi achiev initi success classif relat data text document graph paper conduct systemat explor frequent pattern base classif provid solid reason support methodolog featur combin pattern captur underli semant singl featur inclus infrequ pattern significantli improv accuraci due limit predict power build connect pattern frequenc discrimin measur inform gain fisher score develop strategi set minimum support frequent pattern mine gener pattern base strategi coupl propos featur select algorithm discrimin frequent pattern gener build high qualiti classifi demonstr frequent pattern base classif framework achiev good scalabl high accuraci classifi larg dataset empir studi signific improv classif accuraci achiev uci dataset select discrimin frequent pattern gapprox mine frequent approxim pattern massiv network recent aris larg number graph massiv size complex structur applic biolog network social network web demand power data mine method due inher nois data divers crucial address issu approxim mine pattern potenti interest toler variat paper investig problem mine frequent approxim pattern massiv network propos method call gapprox gapprox find approxim network pattern kei knowledg discoveri applic structur data enrich librari graph mine methodolog introduc techniqu complet redund free strategi explor pattern space face gapprox transform frequent approxim sens anti monoton constraint push deep mine process systemat empir studi real synthet data set show frequent approxim pattern mine worm protein protein interact network biolog interest gapprox effect effici effici discoveri frequent approxim sequenti pattern propos effici algorithm mine frequent approxim sequenti pattern ham distanc model algorithm gain effici adopt break build methodolog breakdown base observ occurr frequent pattern classifi group call strand develop effici algorithm quickli mine strand iter growth build stage strand group form support set approxim pattern identifi salient featur algorithm abil grow frequent pattern iter assembl build block signific size local search fashion avoid increment growth global search achiev greater effici lose complet mine result experiment studi demonstr algorithm effici mine global repeat approxim sequenti pattern miss exist method systemat discoveri function modul context dspecif function annot human genom systemat discoveri function modul context specif function annot human genom graph dbase approach systemat reconstruct human transcript regulatori modul gprune constraint push framework graph pattern mine abstract graph mine applic increasingli strong urg impos user constraint mine result unlik tradit itemset constraint structur constraint densiti diamet graph hard push deep mine process paper give comprehens studi prune properti tradit structur constraint aim reduc pattern search space data search space gener framework call gprune propos incorpor constraint recurs reinforc entir mine process concept pattern insepar data antimonoton propos handl structur constraint uniqu context graph combin prune properti comprehens unifi classif framework structur constraint explor antimonoton context graph pattern mine signific extens classif constraint deepen understand prune properti structur graph constraint support entiti search larg dscale prototyp search engin support entiti search larg dscale prototyp search engin entityrank search entiti directli holist web evolv data rich repositori standard page view current search engin increasingli inadequ wide rang queri task search data entiti phone number paper pdf date todai engin indirectli page entiti page current engin find page individu search directli holist find inform finer granular studi problem entiti search signific departur tradit document retriev focu core challeng rank entiti distil underli conceptu model impress model develop probabilist rank framework entityrank seamlessli integr local global inform rank evalu onlin prototyp tb web corpu show entityrank perform effect graph contain search index set model graph queri graph contain search aim find model graph due wide adopt graph model fast contain search graph data find applic domain comparison tradit graph search retriev graph contain search index characterist examin paper perform systemat studi characterist propos contrast subgraph base index model call cindex contrast subgraph captur structur differ model graph queri graph perfect index due high select redund awar featur select process cindex sort set signific distinct contrast subgraph maxim index capabl show np complet choos set index featur greedi algorithm approxim level optim index ratio take solut base index model extend accommod hierarch index methodolog appli data space cluster sampl techniqu reduc index construct time propos methodolog gener solut contain search index graph data transit relat experiment result real test data show cindex achiev optim prune power contain search workload confirm obviou advantag indic built tradit graph search scenario compress frequent pattern major challeng frequent pattern mine sheer size mine result compress frequent pattern propos cluster frequent pattern tight measur call cluster select repres pattern cluster problem find minimum set repres pattern shown np hard develop greedi method rpglobal rplocal guarante compress bound higher comput complex sacrific theoret bound effici perform studi show compress qualiti rplocal close rpglobal reduc number close frequent pattern order magnitud rplocal mine faster fpclose fast close frequent pattern mine method show rpglobal rplocal combin balanc qualiti effici frequent pattern mine current statu futur direct mine index similar search graph complex structur search substructur superimpos distanc effici index techniqu develop exact approxim substructur search larg scale graph databas retriev problem structur categor geometr distanc constraint solv paper develop method call pi partit base graph index search support similar search substructur superimpos distanc constraint pi select discrimin fragment queri graph index prune graph violat distanc constraint identifi criterion distinguish select fragment multipl graph develop partit method obtain set highli select fragment improv prune perform experiment result show pi effect process real graph queri extract redund dawar top dk pattern mine control flow abnorm logic error isol featur dbase similar search graph structur similar search complex structur import oper graph relat applic exact match restrict articl investig issu substructur similar search index featur graph databas transform edg relax ratio queri graph maximum allow featur miss structur filter algorithm filter graph perform pairwis similar comput shown featur result poor filter perform challeng design effect featur set select strategi maxim filter capabl prove complex optim featur set select worst case number featur select practic identifi criteria build effect featur set filter demonstr combin featur similar size select improv filter search perform significantli multi filter composit framework propos featur base filter concept gener appli search approxim consecut sequenc tree structur data categori subject descriptor databas manag system queri process preliminari releas articl accept acm transact databas system definit version product acm releas supersed version integr arrai analyz softwar packag analysi cross dplatform cross dspeci microarrai data statist debug hypothesi test dbase approach abstract manual debug tediou costli high cost motiv develop fault local techniqu develop search fault locat paper propos statist method call sober automat local softwar fault prior knowledg program semant unlik exist statist approach select predic correl program failur sober model predic evalu correct incorrect execut predic fault relev evalu pattern incorrect execut significantli diverg correct featur rational similar hypothesi test sober quantifi fault relev predic principl systemat evalu sober set previou studi result demonstr effect sober develop locat fault siemen suit examin percent code transit approach propos holger al statist approach liblit al locat fault effect sober evalu imperfect world test suit inadequ partial label experi sober achiev competit qualiti harsh circumst case studi grep bc report shed light applic sober larg program index term debug aid statist method statist debug sober statist model dbase bug local mine coher dens subgraph massiv biolog network function discoveri mine close relat graph connect constraint mine coher dens subgraph massiv biolog network function discoveri summar itemset pattern profil dbase approach frequent pattern mine studi extens scalabl method mine kind pattern includ itemset sequenc graph bottleneck frequent pattern mine effici interpret due huge number pattern gener mine process paper examin summar collect itemset pattern repres small number pattern user handl easili repres cover frequent pattern approxim support gener model built extract profil repres support pattern easili recov consult origin dataset base restor error propos qualiti measur function determin optim paramet polynomi time algorithm develop optim heurist effici improv empir studi obtain compact summar real dataset mine close relat graph connect constraint commun mine multi drelat network abstract social network analysi attract attent recent year commun mine major direct social network analysi exist method commun mine assum kind relat network mine result independ user prefer realiti exist multipl heterogen social network repres kind relationship kind relationship plai distinct role task paper systemat analyz problem mine hidden commun heterogen social network base observ relat import respect queri propos method learn optim linear combin relat meet user expect obtain relat perform achiev commun mine mine behavior graph backtrac noncrash bug analyz execut buggi softwar program essenti data mine process interest method develop trace crash bug memori violat core dump difficult analyz noncrash bug logic error paper develop method classifi structur trace program execut softwar behavior graph analyz correct incorrect execut made good progress isol program region lead faulti execut classif framework built integr close graph mine svm classif interestingli suspici region identifi captur classif accuraci chang measur increment program execut perform studi case base experi show approach effect effici seqindex index sequenc sequenti pattern analysi paper studi issu relat design construct high perform sequenc index structur larg sequenc databas build effect indic method call seqindex propos select indic base analysi discrimin frequent sequenti pattern mine larg sequenc databas analysi lead construct compact effect index structur elimin requir set optim support threshold difficult user provid practic discrimin frequent pattern base index method proven effect base perform studi substructur similar search graph databas advanc databas system face great challeng rais emerg massiv complex structur data bioinformat chem informat applic fundament support need applic effici search complex structur data exact match restrict similar search complex structur vital oper support effici paper investig issu substructur similar search index featur graph databas transform edg relax ratio queri graph maximum allow miss featur structur filter algorithm call grafil filter graph perform pairwis similar comput shown featur result poor filter perform challeng design effect featur set select strategi filter examin effect featur select mechan develop multi filter composit strategi filter distinct complementari subset featur identifi criteria form effect featur set filter demonstr combin featur similar size select improv filter search perform significantli concept present grafil appli search approxim consecut sequenc tree complic structur graphmin structur pattern dmine system larg disk dbase graph databas applic mine compress frequent dpattern set major challeng frequent pattern mine sheer size mine result case high min threshold discov commonsens pattern low gener explos number output pattern sever restrict usag paper studi problem compress frequent pattern set typic frequent pattern cluster tight measur call cluster repres pattern select cluster find minimum set repres pattern np hard develop greedi method rpglobal rplocal guarante compress bound higher comput complex sacrific theoret bound effici perform studi show compress qualiti rplocal close rpglobal reduc number close frequent pattern order magnitud rplocal mine faster fpclose fast close frequentpattern mine method show rpglobal rplocal combin balanc qualiti effici graph index base discrimin frequent structur analysi tsp mine top dk close sequenti pattern sequenti pattern mine studi extens data mine commun previou studi requir specif minimum support threshold perform mine difficult user provid threshold practic overcom difficulti propos altern task mine top frequent close sequenti pattern length desir number close sequenti pattern mine minimum length pattern mine close pattern compact represent frequent pattern incspan increment mine sequenti pattern larg databas real life sequenc databas grow increment undesir mine sequenti pattern scratch time small set sequenc grow sequenc ad databas increment algorithm develop sequenti pattern mine mine adapt increment databas updat nontrivi mine sequenti pattern increment exist sequenc grow increment growth lead gener pattern due interact grow subsequ origin studi develop effici algorithm incspan increment mine sequenti pattern explor interest properti perform studi show incspan outperform previous propos increment algorithm increment wide margin graph index frequent structur dbase approach graph increasingli import model complic structur schemaless data protein chemic compound xml document graph queri desir retriev graph quickli larg databas graph base indic paper investig issu index graph propos solut appli graph mine techniqu di erent exist path base method approach call gindex make frequent substructur basic index featur frequent substructur ideal candid explor intrins characterist data stabl databas updat reduc size index structur techniqu size increas support constraint discrimin fragment introduc perform studi show gindex time smaller index size achiev time perform comparison typic path base method graphgrep gindex approach eleg solut graph index problem demonstr databas index queri process benefit data mine frequent pattern mine concept develop appli index sequenc tree complic structur sequenti pattern mine structur pattern mine pattern dgrowth approach tsp mine top dk close sequenti pattern sequenti pattern mine studi extens data mine commun previou studi requir specif minimum support threshold perform mine difficult user provid threshold practic overcom difficulti propos altern task mine top frequent close sequenti pattern length desir number close sequenti pattern mine minimum length pattern mine close pattern compact represent frequent pattern closegraph mine close frequent graph pattern categori subject descriptor clospan mine close sequenti pattern larg databas gspan graph dbase substructur pattern mine investig approach frequent graph base pattern mine graph dataset propos algorithm call gspan graph base substructur pattern mine discov frequent substructur candid gener gspan build lexicograph order graph map graph uniqu minimum df code canon label base lexicograph order gspan adopt depth search strategi mine frequent connect subgraph effici perform studi show gspan substanti outperform previou algorithm order magnitud 