generalis sequenc signatur symbol cluster mine core pattern stock market data establish relationship pattern stock market data work grant establish concept work divers data step abstract develop appropri repres rich data type sequenc graph data combin convent data type boolean item data step integr divers data techniqu develop establish signific attribut group correspond data sourc find signific relationship attribut group boolean attribut extract pattern graph relat data address abstract graph data character structur graph integr inform graph node develop techniqu determin combin attribut unexpect respect basic properti node data base simpl graph correl fig illustr concept cluster sequenc overlap pattern dbase time dseri subsequ cluster radial distribut function subspac sum extract drandom data massiv nois data mine vector ditem pattern neighborhood histogram pattern mine approach discov gener sequenc signatur pattern mine approach discov gener sequenc signatur typic sequenc signatur motif domain assum local region sequenc deriv combin gener concept sequenc signatur introduc algorithm effici determin signatur base subsequ locat sequenc preprocess step sequenc transform featur space subsequ mine gener signatur evalu signatur relat inter pro databas highlight differ comparison interpro show signatur deriv sequenc annot higher confid relat gene express data dcompon system function annot escherichia coli gener sequenc signatur symbol cluster effici weight nearest neighbour classifi vertic data represent exploit edg semant citat graph effici vertic arm bison bio dinterfac semi dglobal analysi network pattern bison bio interfac semi global analysi network pattern integr profil hidden markov model output associ rule mine differenti associ rule mine studi protein dprotein interact network protein protein interact great interest biologist varieti high throughput techniqu devis lead separ definit interact network concept differenti associ rule mine introduc studi annot protein context interact network differ item edg network explicitli target step identifi differ network separ defin set node techniqu differenti associ rule mine appli comparison protein annot interact network interact network case find rule explain properti protein interact network rule show promis advanc studi decis tree induct dynam high ddimension data dtree decis tree induct power classif tool practic work static data dozen attribut adapt decis tree concept set data rapidli hundr thousand attribut relev decis tree branch evalu need base recent data focus data classifi algorithm base tree data structur fast evalu count data point result scale linear data set size kernel dbase semi dna efv bayesian classifi dtree pine podium increment neighbor evalu classifi spatial data set train data nearest neighbor classif predict class unknown tupl search train set nearest neighbor classifi frequent class neighbor nearest neighbor cast equal vote class paper propos algorithm podium increment neighbor evalu pine nearest neighbor weight vote metric call hobbit distanc metric data structur tree effici implement pine algorithm spatial data experi show gaussian podium function pine outperform nearest neighbor knn method term classif accuraci spatial data addit pine algorithm instanc potenti neighbor prespecifi knn method assign high weight nearest neighbor low weight neighbor high classif accuraci achiev effici hierarch cluster larg data set dtree hierarch cluster method attract attent give user maximum amount flexibl requir paramet choic predetermin result repres level granular paper hierarch method introduc fundament relat partit method medoid mean densiti base method center defin denclu superior mean medoid reduct outlier influenc avoid time complex partit base algorithm storag requir densiti base implement present suit spatial stream multimedia data tree effici data storag access lazi classifi dtree lazi classifi store train sampl build classifi sampl classifi differ eager classifi decis tree induct build gener model decis tree receiv sampl nearest neighbor knn classif typic lazi classifi set train data knearest neighbor classifi predict class unknown tupl search train set nearest neighbor assign common class nearest neighbor lazi classifi faster train time eager classifi slower predic time comput delai time paper introduc approach effici construct lazi classifi data structur peano count tree tree treeisa lossless compress represent origin data record count inform facilit effici data mine tree structur introduc classifi tree base nearest neighbor classifi pknn podium increment neighbor evalu pine perform analysi show algorithm outperform classic knn method dtree classif yeast gene delet data genom data properti make typic relat data presenc multi valu attribut larg number null valu led tree base bit vector represent match valu count evalu similar gene quantit inform number interact includ classifi interact inform allow extend properti protein inform interact neighbor featur attribut weight independ relev attribut systemat evalu optim weight genet algorithm aroc classifi list fit function genet algorithm 