基于自然語言的Apriori關聯規(guī)則的視覺挖掘方法

基于自然語言的Apriori關聯規(guī)則的視覺挖掘方法摘要：抽象-可視化數據挖掘技術可以以圖形方式向用戶展示數據挖掘過程，從而使用戶更易于理解挖掘過程及其結果，而且在數據挖掘中也非常重要。然而，現在大多數視覺數據挖掘都是通過可視化的結果而進行的。同時，它不適用于關聯規(guī)則的可視化處理的圖形顯示。鑒于上述缺點，本文采用自然語言處理方法，以自然語言視覺地進行Apriori關聯規(guī)則的整體挖掘過程，包括數據預處理，挖掘過程和挖掘結果的可視化顯示為用戶提供了一套具有更多感知和更易于理解的特征的集成方案關鍵字：apriori關聯規(guī)則數據挖掘可視化1引言視覺數據挖掘技術是可視化技術和數據挖掘技術的結合。使用計算機圖形、圖像處理技術等方法將數據挖掘的源數據，中間結果和最終挖掘結果轉換成易于理解的圖形或圖像，然后進行貫穿的理論，方法和技術交互式處理。根據數據挖掘應用中可視化的不同階段，數據挖掘的可視化可以分為源數據可視化，挖掘過程可視化和結果可視化。源數據可視化源數據可視化方法在數據挖掘之前，以可視化的形式將整個數據集呈現給用戶。目的是使用戶能夠快速找到有趣的地區(qū)，從而實現挖掘目標和目標的下一步。過程可視化過程可視化實現起來相當復雜。主要有兩種方法-一種是在采礦過程中可視化地呈現中間結果，并使用戶根據中間結果的反饋方便地調整參數和約束。另一種方法是以圖標和流程圖的形式保持整個數據挖掘過程，根據用戶可以觀察數據源，數據集成，清理和預處理過程以及采礦結果的存儲和可視化等等。（3）結果可視化數據挖掘結果可視化是指在采礦過程結束時以圖形和圖像的形式描述挖掘結果或知識，以提高用戶對結果的理解，并使用戶更好地評估和利用采礦結果。2、國外家庭視覺數據挖掘研究狀況目前，視覺數據挖掘技術的研究在國內外都處于起步階段，如何使用可視化技術來顯示利用各種數據挖掘算法生成后的模型。該方向的主要研究內容是通過一些特殊視覺圖形中的關聯規(guī)則、決策樹和聚類等算法向用戶顯示生成的結果，以幫助用戶更好地了解結果數據挖掘模型。典型的業(yè)務應用程序是IBMSPSSModeler，開源工具包括Weka、Orange、GGobi和KNIME，以及GoogleVisualPublicPlatform：PublicDataExplorer。視覺數據挖掘工具是一種很好的數據分析工具，在行業(yè)應用中，使用可視化數據挖掘工具顯示數據挖掘更為明確，結合數據挖掘技術，更有利于分析的數據挖掘結果。目前，關聯規(guī)則的可視化研究主要集中在可視化數據和關聯規(guī)則結果上，而挖掘過程可視化存在很多缺陷。特別是在視覺演示過程中，基本采用圖形形式。在實踐中已經發(fā)現，圖形方法不適合在過程中顯示關聯規(guī)則及其結果。因為對于關聯規(guī)則，我們的目的是找到頻繁的項目集，最好的結果顯示它們是文本，同時對于最終獲得的關聯規(guī)則，圖形應用程序不能夠很好地顯示，最好的方法是用基于自然語言的方式顯示應用程序。本文提出了基于自然語言的Apriori關聯規(guī)則的視覺挖掘方案。該方案的預處理，中間過程和采礦結果各個方面均可視化。旨在通過最可接受的自然語言作為工具，實現整個采礦過程的視覺演示。3基于APRIORI協會規(guī)則的可視化采礦的基本理念本文提出的關聯規(guī)則的視覺挖掘基本思想是在數據挖掘的整個過程中，提前提出關聯規(guī)則的視覺挖掘基本上是關于采礦結果可視化的，很少涉及中間和預處理過程中的可視化。對于結果可視化，圖形方法是主要采用的顯示方式，如使用平行坐標法，有向圖法等。然而，對于關聯規(guī)則，通過頻繁項目集和關聯規(guī)則的方式進行圖形顯示似乎無能為力。協會只是反映規(guī)則，規(guī)則最直接的形式是使用自然語言，而奧術公式和圖形對于那些非常專業(yè)的人員而言是可以理解的，不適合普及。而且，當然，充分運用反映關聯規(guī)則的自然語言對實現有一定困難。在本文中，采用自然語言的形式，以視覺方式展示了整個采礦過程?？梢暬^程如圖1所示圖1關聯規(guī)則的視覺過程表1數學分數變換規(guī)則序號條件等級A1Math>=85優(yōu)A2Math>=60andMath<=85中A3Math<60差（1）數據預處理數據預處理是整個數據挖掘的關鍵，也是第一步，一般程序自動完成工作并顯示差異。本文采用完全互動的預處理操作可視化方法，首先構建用戶定義的自然語言轉換規(guī)則庫，易于編輯規(guī)則，其最終目標是將屬性值轉換為自然語言。例如，表1可以被定義為這樣的規(guī)則，根據得分值，不同的分數可以被轉換成不同的代碼。（2）采礦過程挖掘過程的可視化主要體現在中間挖掘結果的視覺顯示和用戶與系統之間的相互作用。對于關聯規(guī)則，中間挖掘結果體現在頻繁項集合的顯示中，以供用戶觀察采礦過程正確或不正確，同時根據交互程序，用戶可以及時地介入方案進行運作（3）采礦結果挖掘結果可視化主要是基于最大頻繁項集來提取關聯規(guī)則，并通過轉換規(guī)則將編碼關聯規(guī)則轉換為自然語言形式。用戶可以一目了然地了解規(guī)則的含義。4APRIORI協會規(guī)則的視覺采礦實施數據預處理可視化構建轉換表：轉換表（字段名稱，代碼，條件和含義）圖2數據預處理可視化用戶可以在轉換規(guī)則表中進行編輯，包括添加，刪除等。形成轉換規(guī)則表后，從數據預處理開始。如圖2所示，首先打開原始數據表，掃描表中的每個屬性和值，并在轉換規(guī)則表中查找屬性和值，并進行轉換，如果沒有找到相應的屬性和值，然后反復進行錯誤處理，直到轉換完成B視覺挖掘過程1）采礦參數設定在挖掘之前，用戶選擇支持度和置信度，然后開始進行數據挖掘，其中可以隨時觀察采礦頻繁項目集和最大頻繁項集的變化，如果異常，可能會及時終止程序的運行并重新選擇參數以重復數據挖掘。2）中間結果顯示在采礦過程中，可以顯示初始數據項集，頻繁項集，最大頻繁項集，以便觀察用戶數據挖掘的整個過程。C采礦結果可視化1）根據模糊關聯理論建立關聯規(guī)則的模糊運算符規(guī)則有兩個限制，一個是支持度，另一個是置信度。建立關聯規(guī)則的關鍵在于信心度量，因此本文以信度為參照。根據需要，在本文中，置信度取0-100的水平作為邊界，所以模糊理論的領域表達為[0,100]。模糊集的特征函數被稱為隸屬函數，它是描述逐漸變化的東西和“中介轉型”現象的關鍵。下屬功能有很多種，常用的有三種形式：正常型，基于環(huán)型類型和環(huán)型。從經驗來看，建議使用基于環(huán)型類型或環(huán)型的會員功能來描述模糊操作者，而選擇正常類型來描述模糊運算符。運算符是其模糊度的描述，本文表明了關聯規(guī)則的建立程度。我們使用“很可能”，“可能”，“更可能”，“可能”修改關聯規(guī)則的建立程度。其中a為閾值，λ為操作者的對應值，Hλ為定量描述模糊值的操作符。設置A的模糊值，定義Hλ，對于HλA=Aλ，并且λ的值的相應語義含義應該是“很可能”，λ=4;“可能”，λ=2;“更可能”，Aλ=0.5;“可能”，λ=0.25?；谀：阕?，模糊條件由公式（1）得出，通過公式（1）可以推導出精確的范圍：2）關聯規(guī)則的自然語言轉換如圖3所示，為關聯規(guī)則形成以符號形式顯示，掃描轉換表，掃描規(guī)則中的每個符號，將符號轉換為自然語言，最后通過自然語言將符號中的顯示規(guī)則轉換為規(guī)則。例如，符號規(guī)則：B2-->F3轉換成：中等職業(yè)成就-->方向（就業(yè)）表3挖掘結果的可視化過程為了測試本文方法的可行性，根據Apriori關聯規(guī)則挖掘算法，編寫了學生成績與畢業(yè)指導關系的數據挖掘程序。以N大學X學院C學院為例：64名學生，5名綜合表現屬性，1名畢業(yè)方向屬性，挖掘過程和結果如圖4和圖5所示。從圖4和圖5可以看出，它在整個過程中主要建立自然語言轉換規(guī)則庫，然后將屬性值轉換為代碼，并使用代碼進行數據挖掘?？梢杂^察采礦過程中頻繁項集的變化，使用戶能夠及時調整初始參數。挖掘結果可以直接以自然語言顯示，以提高規(guī)則的可讀性。圖4原始數據的預處理圖5數據挖掘結果5結論本文針對目前大多數現有的視覺數據挖掘技術已經集中在數據挖掘結果的可視化這兩個缺陷，同時對于Apriori關聯規(guī)則，其視覺處理不適合圖形顯示，提出了一種基于自然語言的視覺處理方法。該方法可以對關聯規(guī)則的Apriori算法進行數據預處理，并對整個挖掘過程中的挖掘過程和結果進行自然語言視覺處理。它提供了一套具有更多視覺和易于理解的特征的集成方案。擴大了視覺數據挖掘過程的應用范圍，有利于數據挖掘技術的推廣應用。參考文獻：[1]XieQinghua,ZhangNingrong,SongYishenetc,"TheVisualModelMethodandTechnologyofClusteringDataMining",JoumalofPLAuniversityofscienceandtechnology(NaturalScienceEdition),vo1.I6,no.I,pp.7-15,2015.[2]ZhangJun,"ResearehandImplementationofVisualDataMiningTechnology",JoumalofChongqingTechnologyandBusinessUniversity(NaturalScienceEdition),vo1.30,no.3,pp.58-61,95,2013.[3]WangJing,"TheResearchandApplicationofVisualTechnologyinDataMining",JilinUniversitypress,Changchun,2009.[4]HuJun,"TheVisualDataMiningModelandItsApplicationResearch",BeijingJiaotongUniversitypress,Beijing,2009.[5]SongChengzhang,HuangXiaodong,LiPeng,ete,"PublieSentimentLargeDataAnalysisBasedonProcessingandItsVisualizationStudy",FujianComputer,no.5,pp.19-21,2014.[6]LiHuijun,LiZhiquan,"TheResearchontheVisualClusteringMethodBasedonImprovingRadarMap",JournalofYanshanUniversity,vo1.5,no.1,pp.58-62,2013.[7]SunQiunian,RaoYuan,"TheResearchOverviewofNetworkDataVisualizationTechnologyBasedontheCOITelationAnalysis",ComputerSeience,no.6a,pp.484-488,2015.[8]LiYang,HaoZhifeng,XiaoYanShan,ete,"TheMultidimensionalDataVisualizationoftheDifferentialPrivacyDPEk-meansundertheDataAggregation",Small-sizeComputerandMicrocomputerSystem,no.7,pp.1637-1640,2013.[9]HuangBin,XuShuren,PuWei,"TheDesignandImplementationofDataMiningPlatformBasedonMapReduee",ComputerEngineeringandDesign,no.2,pp.495-50I,2013.[10]QiSenyu,DuJinglin,QianShenshen,etc,"TheResearchOverviewofMultidimensionalDataVisualizationTechnology[J].SoftwareGuide2015,14(7):15to17.[11]YangZhenyu,WangXiaoyue,BaiRujiang,"TheComparativeAnalysisResearchonMajorForeignVisualDataMiningOpen-sourceSoftwareS",LibraryTheoryandaetiee,no.5,pp.89-93,2013.[12]ZhangJun,"TheResearchandImplementationofVisualDataMiningTeehnology",JoumalofChongqingTechnologyandBusinessUniversity(NaturalScienceEdition),vol.30,no.30.58-61,2013.[13]DengWenhong,ZhouZhongli,SongZhenming,ete,"TheResearchonDecisionSupportSystemBasedontheFocusVisualization",JoumalofXinyangNormalUniversity(NaturalScienceEdition)voI.26,no.l,pp.128-132,2013.[14]LiZheng,KangLiyuan,FanXiaohui,"TheIntegrationDataMiningandVisualizationTechnologyResearchonTraditionalChineseMedicinePharmaceuticaProcessData",ChineseJournalofTraditionalChineseMedieine,vo1.33,no.l5,pp2989-2992,2014.[15]WangSong,WuYadong,LiQiusheng,ete,"TheWeiboEvolutionVisualizationBasedontheTime-spaceAnalysis",JoumalofSouthwestUniversityofScienceandTeehnology,vo1.29,no.3.pp.68-75,2014.[16]FuSha,ZhouHangjun,"TheResearchandImprovementontheAprioriAlgorithmoftheAssociationRuleMining",MicroeleetronicsandComputer,voI.30,no.9,pp.110-114,2013.TheVisualMiningMethodofAprioriAssociationRuleBasedonNaturalLanguageZhangChunshengCollegeo/ComputerScienceandTechnologyInnerMongoliaUniversityForNationalities,tongliao028043,CHINAzhangcs_817@163.comAbstract-Visualdataminingtechniquescandisplaytheprocessofdataminingandresultstotheusergraphically,whichmakestheusermoreperceptualandeasytounderstandthemeaningoftheminingprocessanditsresultsandmoreoveritisveryimportantindatamining.However,mostofthevisualdataminingnowisprogressedwiththeresultofvisualization.Atthesametime,itisnotsuitableforthegraphicaldisplaytothevisualizationprocessingoftheassociationrule.Inviewoftheaboveshortcomings,inthispaper,thewholeminingprocessofAprioriassociationruleisvisuallyconductedunderthenaturallanguagebythewayofthenaturallanguageprocessingmethod,includingdatapreprocessing,miningprocessandthevisualizationdisplayofminingresultswhichprovidesanintegratesetofschemesfortheuserwithcharacteristicsofbeingmoreperceptualandmoreeasytounderstand.Keywords-apriori;theassociationrule;datamining;visualization(keywords)INTRODUCTIONVisualdataminingtechnologyisthecombinationofvisualizationtechnologyanddataminingtechnology.Itistheverytheory,methodandtechnologytousecomputergraphics,imageprocessingtechnologyandothermethodstotransformthesourcedataofdatamining,theintermediateresultsandthefinalminingresultsintoperceptuallyandeasilyunderstandablegraphicsorimagesandthentocarrythroughinteractiveprocessing.Accordingtodifferentstagesofvisualizationintheapplicationofdatamining,thevisualizationofdataminingcanbedividedintothesourcedatavisualization,themining-processvisualizationandtheresultvisualization[1-7].1)ThesourcedatavisualizationThesourcedatavisualizationmethodispriortodataminingandpresentthewholedatasettotheuserintheformofvisualization.Thepurposeistoenabletheusertoquicklyfindtheinterestingregion,soastoimplementthenextstepofdiggingwithaimandtarget.2)TheprocessvisualizationThemining-processvisualizationisfairlycomplextoimplement.Therearemainlytwoways-oneistovisuallypresenttheintermediateresultsintheprocessofminingandmaketheuserconvenientlyadjustparametersandconstraints

LiYanCollegeo/ComputerScienceandTechnologyInnerMongoliaUniversityForNationalities,tongliao028043,CHINAliyan_yx@126.comaccordingtothefeedbackoftheintermediateresults.Anotherwayistokeepthewholedataminingprocessintheformoficonsandflowchartsandaccordingtothem,theusercanobservethedatasource,thedataintegration,thecleaningandpretreatmentprocess,andthestorageandvisualizationofminingresults,etc.3)TheresultvisualizationThedata-miningresultvisualizationreferstodescribingtheminingresultsorknowledgeintheformofgraphicsandimagesattheendoftheminingprocessinordertoimprovetheuser'sunderstandingoftheresults,andmaketheuserbetterevaluateandmakeuseoftheminingresults.VISUALDATAMININGRESEARCHSTATUSATHOMEANDABROADAtpresent,thestudyofvisualdataminingtechnologyisintheascendantbothathomeandabroad,ofwhichhowtousethevisualizationtechnologytodisplayconsequencemodelsgeneratedwithvariouskindsofdataminingalgorithm.Themainresearchcontentofthisdirectionistodisplaythegeneratedresultstotheuserwithalgorithmsuchasassociationrules,decisiontreeandclusteringinsomespecialvisualgraphicinordertohelptheuserunderstandtheresultdataminingmodelbetter.ThetypicalbusinessapplicationisIBMSPSSModeler,andtheopensourcetoolsareWeka,Orange,GGobiandKNIME,plusGoogleVisualPublicPlatform:PublicDataExplorer.Visualdataminingtoolisagoodkindofdataanalysistoolandinindustryapplication,itismoreexplicittoshowthedataminingwiththeuseofvisualdataminingtool,combinedwiththedataminingtechnologyandalsoitismoreconducivetotheanalysisofthedataminingresults.Atpresentthevisualizationstudyoftheassociationrulearemainlyconcentratedonthevisualizationofdataandtheassociationruleresults[8-16],whiletherearealotofdefectsontheminingprocessvisualization.Especiallyintheprocessofvisualdemonstration,thegraphicformisbasicallyadopted.Ithasbeenfoundinpracticethatthegraphicmethodisnotsuitableforthedisplayoftheassociationruleinthemiddleoftheprocessanditsresults.Becauseasfortheassociationrule,ouraimistofindfrequentitemsets,andthebestresultdisplaythemistext,meanwhileasforthefinally-gainedassociationrule,thegraphicapplication572displayispaleandthebestmethodisthedisplayofapplicationbasedonnaturallanguage.ThispaperproposesavisualminingschemeofAprioriassociationrulebasedonnaturallanguage.Thepretreatment,theintermediateprocess,andminingresultsintheschemearevisualizedinalldimensions.Itaimstorealizethevisualdemonstrationofthewholeminingprocesswiththeapplicationofthemostacceptablenaturallanguageasatool.BASICIDEASOFVISUALMININGBASEDONAPRIORlASSOCIATIONRULEThevisualminingbasicideaontheassociationruleputforwardinthispaperisthroughouttheentireprocessofdatamining.Thevisualminingontheassociationruleputforwardbeforehandwerebasicallyaboutthemining-resultvisualization,verylittlereferringtothevisualizationintheintermediateandpretreatmentprocess.Asfortheresultvisualization,thegraphicmethodisamostlyadoptedwaytodisplay,suchasusingthemethodofparallelcoordinates,thedirectedgraphmethodandetc.Fortheassociationrule,however,graphicaldisplaybythewayoffrequentitemsetsandassociationrulesseemspowerlesswithoutvisualization.Associationrulejustreflectsrulesandforrulesthemostdirectformistousenaturallanguage,whilearcaneformulaeandgraphiccanbeonlyunderstandableforthoseveryprofessionalpersonnelandnotsuitableforpopularization.And,ofcourse,fullyapplicationofnaturallanguagereflectingassociationruleontheimplementationhasthecertaindifficulty.Inthispaper,theformofnaturallanguageisadoptedtodisplaythewholeprocessofminingvisually.ThevisualizationprocessisasshowninFigure1.FigureI.ThevisualprocessofassociationruleTABLE1.MATHEMATICALSCORESTRANSFORMINGRULECodePrerequisiteImplicationAlMath>?85GoodinmathA2Math>?60andMath<?85MediuminmathA3Math<60Poorinmath

1)DatapreprocessingDatapreprocessingisthekeytothewholedatamining,alsothefirststep,andthegeneralprocedureisautomaticallytofinishtheworkandvisualizedifferences.Thispaperadoptsafullyinteractivevisualizationmethodforpretreatmentoperation,andfirsttobuildnaturallanguagetransformationrulesbase,whichisdefinedbytheuser,iseasyforeditingrules,anditsultimategoalistoconverttheattributevalueintonaturallanguage.Forinstance,TableIcanbedefinedassuchrulesthataccordingtoscorevaluesdifferentscorescanbeconvertedintodifferentcodes.2)TheminingprocessThevisualizationofminingprocessofismainlyembodiedinthevisualdisplayoftheintermediateminingresultsandtheinteractionbetweentheuserandthesystem.Forassociationrules,theintermediateminingresultsareembodiedinthedisplayoffrequentitemsetsinordertoservefortheusertoobservetheminingprocesscorrectorincorrect,andatthesametime,accordingtotheinteractiveprogram,theusercanintervenetheprogramoperationtimely.3)TheminingresultTheminingresultvisualizationismainlybasedonmaximumfrequentitemsetstoextractassociationrulesandconvertcodingassociationrulesintonaturallanguageformthroughtransformationrules.Theusercanbeclearataglancetounderstandthemeaningofrules.IV. THEVISUALMININGIMPLEMENTATIONOFAPRIORIASSOCIATIONRULEDatapreprocessingvisualizationBuildthetransitiontable:Thetransitiontable(fieldname,code,conditionandimplication)OutsetNoNo ErrorFigure2.Datapreprocessingvisualization573Theusercaneditonthetransformationruletable,includingadding,deleting,andsoon.Afterformingthetransformationruletable,itistostartwithdatapreprocessing.AsshowninFigure2,firstlyopentheoriginaldatatable,scanroundeachattributeandvaluesinthetable,andlookuptheattributeandvalueinthetransformingruletableandconvert,ifthereisnocheckinthecorrespondingattributesandvalues,thenproceedtheerrorhandlingrepeatedlyuntiltheconversionisfinished.Visualminingprocess1)MiningparameterssettingBeforeexcavation,supportdegreeandconfidencedegreeareselectedbytheuser,andthenbegintodatamining,duringwhichitmayatanytimeobserveminingfrequentitemsetsandthechangeofthemaximumfrequentitemsets,ifabnormal,itmayterminatetheprogramrunninginatimelymannerandreselectparameterstorepeatdatamining.2)TheintermediateresultdisplayDuringtheprocessofmining,itcandisplaytheinitialdataitemsets,frequentitemsets,themaximumfrequentitemsetsinordertoobservethewholeprocessofdataminingfortheuser.Theminingresultvisualization1)ThefuzzymoodoperatorofassociationrulesEstablishedaccordingtothetheoryoffuzzyassociationrule,therearetwoconstraints,oneisthesupportdegreeandtheotheristhedegreeofconfidence.Thekeytothefoundingofassociationruleistheconfidencedegree,thusthispaperadoptsconfidencedegreeasaconditionofitsoperator.Accordingtotheneed,inthispaper,theconfidencedegreetakeslevelsof0-100astheboundary,sothedomainofthefuzzytheoryisexpressedas[0,100].Thecharacteristicfunctionofthefuzzysetisknownasthemembershipfunctionanditisthekeytodescribegradualchangingthingsandthephenomenonof"theintermediarytransition".Thesubordinatefunctionhasmanysortsandtherearethreekindsofformswhicharefrequentlyused:normaltype,ontheringtypeandringtype.Fromserviceexperience,itisadvisabletousemembershipfunctionsofontheringtypeorundertheringtypetodescribethefuzzymoodoperatorwhileitismoreappropriatetochoosethenormaltypetodescribethebluroperator.Themoodoperatoristhedescriptionofitsfuzzydegree,andinthispaperitindicatestheestablishmentdegreeofassociationrule.Weuse"verylikely","probably","morelikely","likely"tomodifytheestablishmentdegreeofassociationrule.Thispaperusesthefollowingform:{lX:2:Y[1+(x:Y)2r'Ji(X)XCj)<Y

Amongthem,isasthresholdvalue,Aisasthecorrespondingvalueofthemoodoperator,andHAisasthemoodoperatortoquantitativelydescribethefuzzyvalue.SetupthefuzzyvalueforA,defineHI-,forHAA=AA,andthecorrespondingsemanticimplicationofthevalueofI-,shouldbe"verylikely",I-,=4;"Probably",A=2;"Morelikely",A=0.5;"Likely",I-,=0.25.Basedonfuzzymoodoperator,thefuzzyconditionisderivedfromtheformula(1),bytheformula(1)itcandeducethepreciserangefor:1/5t1/5t11(2)2)ThenaturallanguageconversionofassociationruleAsshowninFigure3,fortheassociationruleformedbythedisplayinsymbolicform,scanconversiontable,scaneachsymbolintherule,convertsymbolsintonaturallanguage,andfinallyusetherulebythedisplayinsymbolstoconvertintotherulebythedisplayinnaturallanguage.Forinstance,symbolrule:B2--+F3convertinto:mediumprofessionalachievementin--+direction(employment)ScanningtheassociationruleQueryfortherulebaseTransformingintonaturallanguagNoFigure3.ThevisualizationprocessofminingresultTHEINSTANCESTotestthefeasibilityofthemethodinthispaper,writeadataminingprocedureoftherelationbetweenstudents'achievementsandgraduationdirectionsbasedonthealgorithmofAprioriassociationrulemining.TakeXHanClassofCCollegeinNuniversityasanexample:64students,5comprehensiveperformanceattributes,1graduationdirectionattribute,theminingprocessandresultareshowninFigure4andFigure5.AsyoucanseefromFigure4andFigure5,itinthewholeprocessprimarilyestablishthenaturallanguagetransformationrulebase,thenconvertattributevaluesintocodes,andusecodetoconductdatamining.Itcanobservechangesoffrequentitemsetsintheminingprocessandmakesiteasyfortheusertotimelyadjustinitialparameters.Theminingresultscanbedirectlydisplayedinnaturallanguagetoimprovethereadabilityoftherules.574VI. THECONCLUSIONThispaper,inviewofsuchtwodefectsthatmostpresentvisualdataminingtechnologieshavefocusedonthevisualizationofdataminingresults,atthesametime,forAprioriassociationruleitsvisualprocessingisnotsuitableforgraphicaldisplay,proposesavisualprocessingmethodbasedonnaturallanguage.ThismethodcandodatapreprocessingontheApriorialgorithmofassociationrule,anddoanaturallanguagevisualhandlingontheminingprocessandresultsduringthewholeminingprocess.Itprovidesanintegratesetofschemewhichhasthecharacteristicsofmorebeingvisualandeasytounderstand.Itexpandstheapplicationrangeofthevisualdataminingprocessandisconduciveforthepromotionandapplicationofdataminingtechnology.Figure4.ThepreprocessingoftheoriginaldataFigure5.Thedataminingresult

REFERENCESXieQinghua,ZhangNingrong,SongYishenetc,"TheVisualModelMethodandTechnologyofClusteringDataMining",JoumalofPLAuniversityofscienceandtechnology(NaturalScienceEdition),vo1.I6,no.I,pp.7-15,2015.ZhangJun,"ResearehandImplementationofVisualDataMiningTechnology",JoumalofChongqingTechnologyandBusinessUniversity(NaturalScienceEdition),vo1.30,no.3,pp.58-61,95,2013.WangJing,"TheResearchandApplicationofVisualTechnologyinDataMining",JilinUnive