classif induct rule learn negat featur abstract paper report investig compar number strategi includ negat featur process induct rule learn irl emphasi gener negat featur rule learnt includ deriv negat featur part input strategi consid base manipul featur space comparison made associ rule learn arl context multi class text classif result option includ negat featur irl process produc effect classifi kei word rule learn negat multi class text classif frequent pattern trend analysi social network abstract paper describ approach identifi compar frequent pattern trend social network frequent pattern trend defin sequenc time stamp occurr support valu specif frequent pattern exist data trend gener epoch trend sequenc epoch identifi case great trend identifi difficult interpret result combin constraint frequent pattern cluster cluster analysi techniqu argu analysi result enhanc cluster techniqu organis map approach produc sequenc map epoch map compar movement trend identifi frequent pattern trend mine framework evalu standard type social network cattl movement network insur quot network find frequent subgraph longitudin social network data weight graph mine approach cluster configur metric multiag base cluster region interest base imag categor multi dagent base cluster gener multi dagent data mine abstract framework multi agent data mine madm framework compris collect agent cooper address data mine task fundament concept underpin framework support gener data mine vision system grow organ manner central issu facilit growth commun medium requir support agent interact issu partli address natur propos architectur partli extend ontolog advantag offer framework illustr paper cluster applic motiv cluster algorithm identifi agent base approach adopt identifi cluster applic serv demonstr full potenti madm keyword multi agent data mine agent base cluster frequent dgraph mine edg weight graph abstract frequent graph mine entail signific overhead concern candid set gener isomorph check issu respect form frequent pattern mine exacerb context frequent graph mine reduc search space address twin overhead weight approach graph mine propos signific issu weight graph mine antimonoton properti typic control candid set gener longer hold paper examin number edg weight scheme suggest strategi control candid set gener strategi incorpor weight variat gspan atw gspan aw gspan ubw gspan complet evalu approach present multi dagent base cluster gener multi dagent data mine abstract framework multi agent data mine madm framework compris collect agent cooper address data mine task fundament concept underpin framework support gener data mine vision system grow organ manner central issu facilit growth commun medium requir support agent interact issu partli address natur propos architectur partli extend ontolog advantag offer framework illustr paper cluster applic motiv cluster algorithm identifi agent base approach adopt identifi cluster applic serv demonstr full potenti madm keyword multi agent data mine agent base cluster hybrid diaaf fr statist textual featur select languag dindepend text classif find tempor pattern noisi longitudin data studi diabet retinopathi trend mine social network studi larg cattl movement databas web dsite boundari detect abstract defin boundari web site archiv inform retriev purpos import complic task paper web page cluster approach boundari detect suggest princip issu featur select hamper observ clear understand web site paper propos definit web site found principl user intent direct boundari detect problem report sequenc experi number cluster techniqu wide rang featur combin featur identifi web site boundari preliminari result report gener combin featur produc result keyword web site definit web page cluster web archiv detect tempor pattern cluster social network studi focus uk cattl movement databas detect tempor pattern cluster social network studi focus uk cattl movement databas rule learn negat issu effect abstract investig rule learn process inclus negat featur object establish negat induct rule learn system effect respect classif paper seek answer question issu relev system featur identif rule refin synthet real dataset illustr solut identifi issu demonstr neg featur induct rule learn system benefici text classif graph mine dbase featur extract abstract graph base approach document classif paper graph represent offer advantag express document encod standard bag word phrase approach improv classif accuraci document set repres graph set weight graph mine algorithm appli extract frequent subgraph process produc featur vector document classif weight subgraph mine ensur classif effect comput effici signific subgraph extract approach valid evalu popular classif algorithm real world textual data set result demonstr approach outperform exist text classif algorithm dataset size dataset increas process extract frequent featur essenti slide window base dual support framework discov emerg trend tempor data dsat algorithm techniqu discov jump emerg pattern jep time seri data slide window techniqu approach effect perform trend analysi order explor itemset variat time propos framework previou work jep reli itemset border constrain search space dsat exploit previous mine time stamp data slide window concept requir memori minimum comput cost low dataset access dsat discov jep na ve approach utilis memori scale linearli larg dataset set demonstr experiment section corpu callosum mr imag classif abstract approach classifi magnet reson mr imag data specif applic classif mri scan data natur corpu callosum approach gener applic variat spectral segment multi scale graph decomposit mechan introduc result segment store quad tree data structur weight variat develop author gspan algorithm appli identifi frequent tree result imag express set frequent tree great decis tree base featur reduct techniqu appli classif take place result show propos approach perform effici effect obtain classif accuraci case applic hybrid statist data pre dprocess approach languag dindepend text classif data pre process import topic text classif tc aim convert origin textual data data mine readi structur signific text featur serv differenti textcategori identifi broadli speak textual data pre process techniqu divid group linguist ii statist iii hybrid amp ii regard languag independ tc studi relat statist aspect natur textual data pre process includ multi dparti argument experi abstract framework pisa conduct dialogu resolv disput correct categoris case unlik previou system conduct dialogu typic involv agent pisa number agent part facilit discuss case permit categor featur framework agent argu directli individu repositori experi previous engin knowledg base usual case knowledg engin bottleneck avoid argument experi enabl real time data mine conduct individu agent find reason support viewpoint critiqu argument parti multiparti dialogu rais number signific issu necessit design choic paper describ issu resolv protocol implement pisa illustr oper pisa base dataset relat nurseri provis final experi compar pisa argu experi classifi noisi data abstract process base argument theori classifi noisi data specif process found concept call argu experi softwar agent argu classif individu case base previous classifi exampl argu experi protocol padua appli binari classif problem pisa appli multi class problem evalu padua pisa oper equal effect classif system absenc nois system perform compar system noisi data keyword classif argument noisi data gener extend multi dagent data mine framework abstract gener extend multi agent data mine madm integr data mine agent base model simul abstract paper introduc integr studi combin data mine dm agent base model simul abm studi paradigm dm abm concern approach appli dm techniqu abm investig invers ii util abm result dm research detail descript approach present paper conclus futur work integr studi end applic classif associ rule mine mammalian mesenchym stem cell differenti abstract paper data mine analyz differenti mammalian mesenchym stem cell msc databas compris kei paramet influenc destini mammalian msc construct paper introduc classif associ rule mine carm data mine techniqu domain tissu engin initi promis research field experiment result show propos approach perform respect accuraci classif predict found rule mine construct msc databas meaning text classif graph mine dbase featur extract abstract graph base approach document classif paper graph represent offer advantag express document encod standard bag word phrase approach improv classif accuraci document set repres graph set weight graph mine algorithm appli extract frequent subgraph process produc featur vector document classif weight subgraph mine ensur classif effect comput effici signific subgraph extract approach valid evalu popular classif algorithm real world textual data set result demonstr approach outperform exist text classif algorithm dataset size dataset increas process extract frequent featur essenti corpu callosum mr imag classif abstract approach classifi magnet reson mr imag data specif applic classif mri scan data natur corpu callosum approach gener applic variat spectral segment multi scale graph decomposit mechan introduc result segment store quad tree data structur weight variat develop author gspan algorithm appli identifi frequent tree result imag express set frequent tree great decis tree base featur reduct techniqu appli classif take place result show propos approach perform effici effect obtain classif accuraci case applic slide window base dual support framework discov emerg trend tempor data dsat algorithm techniqu discov jump emerg pattern jep time seri data slide window techniqu approach effect perform trend analysi order explor itemset variat time propos framework previou work jep reli itemset border constrain search space dsat exploit previous mine time stamp data slide window concept requir memori minimum comput cost low dataset access dsat discov jep na ve approach utilis memori scale linearli larg dataset set demonstr experiment section padua protocol argument dialogu associ rule analysi mesenchym stem cell differenti vitro classif associ rule mine emad extend multi dagent data miner paper describ emad extend multi agent data mine system emad vision commun data mine agent contribut individu interact decentralis control address data mine request emad end user applic research tool paper detail emad vision conceptu framework current implement emad appli data mine task studi sake breviti concentr agent base data classif full descript emad present improv method extract frequent itemset interim dsupport tree research develop intellig system xxiv multiset refer bag set collect element order import element uniqu vase blue red marbl multiset propos class distanc measur metric design multiset recurr theme data mine applic instanc class origin document dbase extract singl dlabel text classif abstract text mine applic investig text classif tc requir experi perform common textcollect result compar altern approach regard singl label tc text collect textual data sourc origin form limit volum textual data larg eas experiment predefin class class consist document document label singl class multipl class document found actual text content paper propos standard approach automat extract qualifi document base textual data sourc effect reliabl singl label tc experi experiment result demonstr document base extract base approach effect singl label tc experi weight util framework mine associ rule associ rule mine arm identifi frequent itemset databas gener associ rule assum item signific frequenc occurr record weight util weight util case item aspect number real applic retail market nutrit pattern mine differ item strong impact decis make applic unlik standard arm framework weight util arm wuarm consid vari signific frequenc valu individu item weight util weight util mine focus identifi itemset weight util higher user weight util threshold conduct experi synthet real data set standard arm weight arm weight util arm wuarm present analysi result extract nutrit pattern np fuzzi associ rule mine abstract paper propos framework mine market basket data gener nutrit pattern np method analys gener nutrit pattern fuzzi associ rule mine edibl attribut filter transact input data project convert requir daili allow rda numer valu rda databas convert fuzzi databas expend normal fuzzi attribut compris fuzzi set analysi nutrit inform perform normal gener associ rule convert fuzzi transact databas approach prototyp support tool extract nutrit pattern np signifi level nutrit content associ rule item paper present perform test interesting measur demonstr effect approach conclud experiment result discuss evalu propos framework weight associ rule mine binari fuzzi data abstract approach present mine weight associ rule ar binari fuzzi data address issu invalid downward closur properti dcp weight associ rule mine item assign weight signific user defin criteria work weight associ rule mine struggl invalid downward closur properti assumpt made valid properti gener weight associ rule mine problem databas binari quantit attribut weight set methodolog apriori approach avoid pre post process oppos weight associ rule mine algorithm elimin extra step rule gener paper conclud experiment result discuss evalu propos approach mine alloc pattern dsum weight item associ rule ar common knowledg model data mine describ implic cooccur relationship disjoint set binari valu transact databas attribut item express form anteced consequ rule variant ar weight associ rule war regard market context paper introduc knowledg model data mine alloc pattern alp alp special form war rule item weight score sum rule item score implic occur relationship disjoint set item weight set inform alloc relationship rule item alp demonstr applic market possibl surpris varieti area propos apriori base algorithm extract hidden interest alp sum weight transact databas experiment result show effect propos algorithm agent base frequent set meta mine introduc emad mine system support dynam creation commun data mine agent explor capabl agent demonstr experi applic data mine distribut data emad restrict data mine task studi sake breviti concentr agent base associ rule mine arm refer frequent set meta mine meta arm full descript propos meta arm model present describ concept meta arm describ analys number potenti solut context emad experiment result consid term number data sourc number record data set number attribut repres mine fuzzi associ rule composit item abstract framework mine fuzzi associ rule ar relat properti composit attribut attribut item featur number valu deriv common schema appli fuzzi associ rule mine arm partit properti valu fuzzi properti set paper describ process deriv fuzzi set composit fuzzi arm cfarm ii uniqu properti arm algorithm found correl factor interesting measur paper includ complet analysi demonstr potenti fuzzi properti ar ii succinct set properti ar gener fuzzi method produc propos approach fuzzi weight associ rule mine weight support confid framework paper extend problem mine weight associ rule classic model boolean fuzzi quantit associ rule mine adopt address issu invalid downward closur properti dcp weight associ rule mine item assign weight signific user defin criteria work dcp struggl invalid downward closur properti assumpt made valid properti gener problem downward closur properti propos fuzzi weight support confid framework boolean quantit item weight set problem invalid dcp solv improv model weight support confid framework classic fuzzi associ rule mine methodolog apriori algorithm approach avoid pre post process oppos weight arm algorithm elimin extra step rule gener paper conclud experiment result discuss evalu propos framework argument base moder benefit assess fuzzi weight associ rule mine weight support confid framework paper extend problem mine weight associ rule classic model boolean fuzzi quantit associ rule mine adopt address issu invalid downward closur properti dcp weight associ rule mine item assign weight signific user defin criteria work dcp struggl invalid downward closur properti assumpt made valid properti gener problem downward closur properti propos fuzzi weight support confid framework boolean quantit item weight set problem invalid dcp solv improv model weight support confid framework classic fuzzi associ rule mine methodolog apriori algorithm approach avoid pre post process oppos weight arm algorithm elimin extra step rule gener paper conclud experiment result discuss evalu propos framework framework mine fuzzi associ rule composit item abstract framework mine fuzzi associ rule ar relat properti composit attribut attribut item featur number valu deriv common schema appli fuzzi associ rule mine arm partit properti valu fuzzi properti set paper describ process deriv fuzzi set composit fuzzi arm cfarm ii uniqu properti arm algorithm found correl factor interesting measur paper includ complet analysi demonstr potenti fuzzi properti ar ii succinct set properti ar gener fuzzi method produc propos approach mine effici signific classif associ rule hybrid rule order classif associ rule mine abstract classif associ rule mine carm approach classifi gener build associ rule mine base classifi classif associ rule car carm algorithm similar set car gener data classifi present order list car base select rule order strategi produc accur classifi essenti develop ration rule order mechan past decad number rule order strategi introduc major identifi confid support amp size anteced csa size padua protocol strategi tactic rule weight approach classif associ rule mine rule order approach classif associ rule mine abstract classif associ rule car common type mine knowledg data mine describ implic occur relationship set binari valu data attribut item pre defin class express form anteced consequ class rule classif associ rule mine carm recent classif rule mine crm approach build associ rule mine arm base classifi car methodolog build classifi present order car list base appli rule order strategi exist rule order mechan identifi confid support size anteced csa size anteced confid support ac weight rel accuraci wra laplac accuraci test paper divid mechan group pure support confid framework ii addit score assign propos hybrid rule order approach combin approach approach ii experiment result show propos rule order approach perform respect accuraci classif statist identif kei phrase text classif abstract algorithm text classif gener involv stage aim identifi textual element word phrase relev classif process stage involv analysi text languag specif possibl domain specif computation costli paper examin number altern keyword gener method phrase construct strategi identifi kei word phrase simpl languag independ statist properti present result demonstr method produc good classif accuraci result obtain phrase base approach keyword text classif text preprocess frequent set meta mine multi dagent data mine dynam rule mine argument base system argument prove influenti reason mechan context multi agent system paper introduc padua protocol argument dialogu associ rule argument formal dynam mine associ rule ar case background mean gener argument exchang dialogu particip ii repres particip background domain knowledg avoid tradit knowledg base represent dialogu particip mine ar case data rule argument counter argument paper fulli describ padua formal propos suit dynam arm algorithm provid support argument process effect threshold valu associ rule base classif accuraci classif associ rule mine carm system oper appli associ rule mine arm method obtain classif rule train ing set previous classifi data rule gener influenc choic arm paramet emploi algorithm typic support confid threshold valu paper examin effect choic predict accuraci carm method show accuraci improv suitabl choic paramet describ hill climb method find paramet set demonstr propos hill climb method effect coupl fast carm algorithm tfpc algorithm tree dbase partit date associ rule mine knowledg bazaar result gener dissatisfact transmiss model learn theori cognit preclud context amp quot constructiv amp quot accur constructivist metatheori drawn sourc domin contemporari educ psycholog constructiv princip implic sack solowai weingrad amp quot activ knowledg product scienc reproduct educ convinc recruit encultur short constructivist analysi knowledg foreground rhetor power persuas difficulti disput amp quot rhetoric concept knowledg construct argu adequ reflect work educ technologist work constructivist theori subtitl rhetor approach constructivist learn realiti check prototyp educ softwar develop embodi lesson knowledg negoti offer educ educ technologist metaphor educ knowledg negoti realiti check prompt educ student independ construct represent task identifi relev variabl give reason evid permit fruit interrog set altern represent realiti check explicitli invit teacher student particip negoti knowledg evid exchang posit clarifi parti educ activ understand importantli make product understand partit strategi distribut associ rule mine paper number altern strategi distribut parallel associ rule mine investig method examin make data structur tree introduc previous author structur organis set attribut support count approach repres wai parallelis basic apriori algorithm method focu mechan partit data process reduc messag pass overhead horizont data distribut vertic candid distribut partit strategi consid includ vertic partit algorithm data vp develop exploit structur tree present experiment result examin perform method implement javaspac conclud javaspac environ candid distribut strategi offer perform distribut origin dataset lower messag overhead data vp algorithm produc result encourag research develop intellig system xxi obtain paramet valu accur classif paper examin effect choic support confid threshold accuraci classifi obtain classif associ rule mine show accuraci improv suitabl choic threshold valu describ method find valu present result demonstr approach obtain higher accuraci coverag analysi train data threshold tune improv classif associ rule mine abstract applic associ rule mine arm identifi classif associ rule car classifi futur instanc popul data mine carm method mine data candid rule prune coverag analysi train data paper describ carm algorithm avoid coverag analysi techniqu tune threshold paramet obtain accur classif present result show approach achiev accuraci compar altern lower cost knowledg bazaar result gener dissatisfact transmiss model learn theori cognit preclud context amp quot constructiv amp quot accur constructivist metatheori drawn sourc domin contemporari educ psycholog constructiv princip implic sack solowai weingrad amp quot activ knowledg product scienc reproduct educ convinc recruit encultur short constructivist analysi knowledg foreground rhetor power persuas difficulti disput amp quot rhetoric concept knowledg construct argu adequ reflect work educ technologist work constructivist theori subtitl rhetor approach constructivist learn realiti check prototyp educ softwar develop embodi lesson knowledg negoti offer educ educ technologist metaphor educ knowledg negoti realiti check prompt educ student independ construct represent task identifi relev variabl give reason evid permit fruit interrog set altern represent realiti check explicitli invit teacher student particip negoti knowledg evid exchang posit clarifi parti educ activ understand importantli make product understand tree partit method memori manag associ rule mine evalu approach classif rule select tree structur mine associ rule data structur associ rule mine dtree dtree tree algorithm author demonstr structur algorithm offer signific advantag term storag execut time dtree vertic partit distribut associ rule mine set attribut weight dnn base binari classif quadrat program abstract nearest neighbour nn method typic lazi learn paradigm solv classif problem method origin propos parameteris method attribut weight set commonli adopt deal irrelev attribut paper propos attribut weight set method nn base classifi quadrat program suitabl binari classif problem method formalis attribut weight set problem quadrat program problem exploit commerci softwar calcul attribut weight evalu method carri seri experi establish data set experi show method practic problem achiev stabl increas accuraci standard nn method competit perform merit method small train set attribut weight set method dnn base binari classif quadrat program abstract paper propos attribut weight set method nn base classifi quadrat program suitabl binari classif problem method formalis attribut weight set problem quadrat program problem exploit commerci softwar calcul attribut weight experi show method practic problem achiev competit perform merit method small train set experiment studi increas divers case dbase diagnosi abstract increas divers case base reason cbr issu recent drawn attent research cbr field diversif techniqu propos discuss literatur extent techniqu bring benefit end user remain question paper report experi appli diversif techniqu case base diagnosi tool product mainten domain result offer evid support diversif techniqu find associ rule frequent attribut formalis optim featur weight set case base diagnosi linear program problem architectur web dbase post dsale servic flexibl manufactur environ comput associ rule partial total problem extract associ rule binari databas exist method involv multipl pass databas cope badli dens pack databas record combinatori explos number set attribut consid present class method singl pass databas perform partial comput total requir store partial total structur facilit comput final total describ gener algorithm comput total storag perform characterist examin method open number strategi practic implement heurist discuss keyword associ rule partial support set enumer introduct establish approach knowledg discoveri databas kdd involv identif associ rule verif show languag np probablist verifi check membership proof logarithm number random bit examin constant number bit proof string languag exist proof verifi accept probabl choic random string string languag verifi reject provid proof probabl result build improv recent result arora safra verifi examin nonconst number bit proof number slowli grow function input length consequ prove max snp hard problem polynomi time approxim scheme np class max snp defin papadimitri yannakaki hard problem class includ vertex cover maximum satisfi maximum cut metric tsp steiner tree shortest superstr improv cliqu hard result feig goldwass lov sz safra szegedi arora safra show exist posit approxim maximum cliqu size vertex graph factor np hard experi discov associ rule legal domain paper explor applic algorithm design find associ rule larg databas discoveri relev associ larg case base tesser spatio dtempor reason multi ddimension data gener applic approach dimension spatial reason approach found uniqu represent base idea tesser address offer comput advantag includ minim data storag computation effici translat data simpl data comparison number dimens consider represent spatial attribut object express simpli concis term set address relat standard set oper express constraint approach incorpor spatial reason system sparta spatial reason tesser address system successfulli conjunct signific number spatial applic domain keyword spatio tempor reason tesser address dimension inform process introduct versatil gener applicabl valid verif knowledg base system theori verif show languag np probablist verifi check membership proof logarithm number random bit examin constant number bit proof string languag exist proof verifi accept probabl choic random string string languag verifi reject provid proof probabl result build improv recent result arora safra verifi examin nonconst number bit proof number slowli grow function input length consequ prove max snp hard problem polynomi time approxim scheme np class max snp defin papadimitri yannakaki hard problem class includ vertex cover maximum satisfi maximum cut metric tsp steiner tree shortest superstr improv cliqu hard result feig goldwass lov sz safra szegedi arora safra show exist posit approxim maximum cliqu size vertex graph factor np hard region descript compar analysi tesser represent rulebas check spatial represent paper describ approach rule base amp found dimension tesser represent space advantag offer represent simpl concis lend computation effect manipul object defin represent standard integ addit subtract encapsul rulebas tesser represent multi dimension arena mechan applic spatial algorithm address rulebas verif algorithm make full advantag represent offer signific reduct time complex tradit approach rulebas verif paper demonstr potenti tesser encod rulebas full impact author opinion explor introduct paper approach rulebas verif val report st intern workshop valid rulebas check spatial represent paper describ approach rule base amp found dimension tesser represent space advantag offer represent simpl concis lend computation effect manipul object defin represent standard integ addit subtract encapsul rulebas tesser represent multi dimension arena mechan applic spatial algorithm address rulebas verif algorithm make full advantag represent offer signific reduct time complex tradit approach rulebas verif paper demonstr potenti tesser encod rulebas full impact author opinion explor introduct paper approach rulebas verif val kd fm knowledg discoveri facil manag databas spatial reason quad tesser represent tesser approach ddimension spatial reason verif valid rulebas binari encod incid matrix techniqu ontolog linear spatial reason ontolog spatial reason base tesser represent space present princip advantag offer represent effect linearis multi dimension space support translat space direct number dimens multi dimension spatial reason implement dimension tempor reason techniqu result concern convent multi dimension spatial reason system base tradit represent longer appli keyword multi dimension spatial reason ontolog introduct ontolog present multi dimension spatial reason base address system effect linearis space number dimens consider result multi dimension reason implement term singl dimens con tempor reason tesser address intellig environment impact assess system spatial reason gi tesser data represent develop distribut databas applic tsl advanc binari encod matrix represent rule base verif repres visual condit legal knowledg base system aspect valid verif knowledg dbase system build knowledg base system maintain electron chart represent interact mainten legal knowledg base system main problem eld arti cial intellig law main problem mccarti construct adequ represent legal graphic interact tool kb mainten aspect valid verif knowledg base system 