|
|
重庆邮电大学学报(自然科学版) 2007
Research on spatial association rules mining in two directionKeywords: data,mining,spatial,data,association,rule,vertical,and,horizontal,direction, Abstract: IndataminingfromtransactionDB,therelationshipsbetweentheattributeshavebeenfocused,buttherelationshipsbetweenthetupleshavenotbeentakenintoaccount.Inspatialdatabase,therearerelationshipsbetweentheattributesandthetuples,andmostoftheassociationsoccurbetweenthetuples,suchasadjacent,intersection,overlapandothertopologicalrelationships.Sothetasksofspatialdataassociationrulesminingincludeminingtherelationshipsbetweenattributesofspatialobjects,whicharecalledasverticaldirectionDM,andtherelationshipsbetweenthetuples,whicharecalledashorizontaldirectionDM.Thispaperanalyzesthestoragemodelsofspatialdata,usesforreferencethetechnologiesofdataminingintransactionDB,definesthespatialdataassociationrule,includingverticaldirectionassociationrule,horizontaldirectionassociationruleandtwodirectionassociationrule,discussesthemeasurementofspatialassociationruleinterestingness,andputsforwardtheworkflowsofspatialassociationruledatamining.Duringtwodirectionspatialassociationrulesmining,analgorithmisproposedtogetnonspatialitemsets.Byvirtueofspatialanalysis,thespatialrelationsweretransferredintononspatialassociationsandthenonspatialitemsetsweregotten.Basedonthenonspatialitemsets,theApriorialgorithmorotheralgorithmscouldbeusedtogetthefrequentitemsetsandthenthespatialassociationrulescomeintobeing.UsingspatialDB,thespatialassociationrulesweregottentovalidatethealgorithm,andthetestresultsshowthatthisalgorithmisefficientandcanminetheinterestingspatialrules.
|
