简介:Fuzzyassociationrules(FARs)canbepowerfulinassessingregionalsoilquality,acriticalsteppriortolandplanningandutilization;however,traditionalFARsminedfromsoilqualitydatabase,ignoringtheimportancevariabilityoftherules,canberedundantandfarfromoptimal.Inthisstudy,wedevelopedamethodapplyingdifferentweightstotraditionalFARstoimproveaccuracyofsoilqualityassessment.AftertheFARsforsoilqualityassessmentweremined,redundantruleswereeliminatedaccordingtowhethertherulesweresignificantornotinreducingthecomplexityofthesoilqualityassessmentmodelsandinimprovingthecomprehensibilityofFARs.Theglobalweights,eachrepresentingtheimportanceofaFARinsoilqualityassessment,werethenintroducedandrefinedusingagradientdescentoptimizationmethod.ThismethodwasappliedtotheassessmentofsoilresourcesconditionsinGuangdongProvince,China.Thenewapproachhadanaccuracyof87%,when15rulesweremined,ascomparedwith76%fromthetraditionalapproach.Theaccuracyincreasedto96%when32rulesweremined,incontrastto88%fromthetraditionalapproach.TheseresultsdemonstratedanimprovedcomprehensibilityofFARsandahighaccuracyoftheproposedmethod.