Treść książki
Przejdź do opcji czytnikaPrzejdź do nawigacjiPrzejdź do informacjiPrzejdź do stopki
11
anaturalcontinuationoftheresearchonoptimizingsequencesoffrequent
itemsetqueries,introducinganotherlevelatwhichprocessingoffrequent
itemsetqueriescanbeoptimized.Weclaimthatadataminingsystemprovided
withasetoffrequentitemsetqueriesatoncehasmorepossibilitiesofsharing
computationsamongthequeriesthanifthequeriesweresubmittedonebyone,
formingasequenceofqueries.Itshouldbenotedthatthegeneralideaof
processingasetofqueriestogetherisnotnewandhasbeenstudiedinthearea
ofdatabasesystemsunderthenameofmulti-queryoptimization.Thus,our
researchcanbeconsideredasyetanotherexampleofpostulatedevolutionof
dataminingsystemsbyborrowingideasandadaptingwell-knownsolutions
fromdatabasesystems.
Weenvisionapplicationsofourtechniquesmainlyforfrequentitemset
queriessubmittedtothesysteminabatchmodeorinresponsetorefreshing
materializedresultsofseveralfrequentitemsetminingtasks.Nevertheless,sets
offrequentitemsetqueriescanbecollectedforexecutionalsoduringusers’
interactivesessionswiththedataminingsystembygroupingthequeries
submittedwithinaconfigurabletimewindow.
1.2.AimandScopeoftheDissertation
Theaimofthisdissertationisdevelopmentoftechniquesofefficient
processingofsetsoffrequentitemsetqueries.Wepresentseveralstrategiesto
addresstheaforementionedresearchproblem,representedbymethods
independentofaparticularfrequentitemsetminingalgorithmaswellastheones
dedicatedtothealgorithmsthatweclaimarerepresentativesofmajor
approachestofrequentitemsetmining.Weshowthataproperlydesigned
processingschemecanbenefitfromsimilaritiesamongthequeriestobe
executedandintypicalscenariosoutperformsequentialindependentexecution,
whichisanaturalreferencepointforassessmentofthequalityofoptimized
batchprocessingmethods.Sinceduetothenatureoffrequentitemsetminingit
isimpossibletoanalyticallyprovesuperiorityofonemethodoveranother,we
discussstrengthandweaknessesofeachproposedtechniqueandverifytheir
efficiencyundervariouscircumstancesexperimentally,followingthewidely
acceptedpracticewithinfrequentpatternminingresearchdomain.Thedetailed
contributionsofthisdissertationinclude:
1)formulationofageneralmodeloffrequentitemsetqueriesindependent
ofparticularlanguagesandinterfaces;
2)formulationoftheproblemofprocessingsetsoffrequentitemsetqueries
andthemodelofdatasharingforbatchesofqueries;
3)threetechniquesofprocessingsetsoffrequentitemsetqueries
independentofaparticularfrequentitemsetminingalgorithm;
4)twotechniquesofprocessingsetsoffrequentitemsetqueriesdedicatedto
theApriorialgorithm;