Proceedings of the International Conference Rough Sets and Emerging Intelligent Systems Par
- 格式:pdf
- 大小:1.23 MB
- 文档页数:10
WS-BPEL研究综述1周进刚,纪勇,王伟东软软件股份有限公司基础软件事业部,辽宁大连(DS006850)E-mail:zhou-jg@摘要:随着Web服务技术的成熟和SOA(面向服务的架构)的研究与应用,WS-BPEL(Web Services Business Process Execution Language,业务流程执行语言)成为面向服务计算(SOC)领域研究的热点,并成为SOA应用中的核心技术。
围绕BPEL,人们开展了多方面的理论与应用研究。
本文首先探讨了Web服务堆栈中与BPEL密切相关的标准技术,然后从工作流相关模式、形式化建模与验证、BPM(Business Process Modeling)领域中BPEL与其它语言模式的比较、转换和集成、面向服务的计算、网格计算和自治计算(Autonomic Computing)等几方面对近年来和当前BPEL相关的领域研究进行了综述,最后对BPEL的发展做了展望。
关键词:BPEL,Web服务,BPM,SOA,面向服务计算中图分类号:TP3111引言WS-BPEL(Web Services Business Process Execution Language,业务流程执行语言,以下简称BPEL)是一种从工业界诞生的标准,由于受业界主流服务及技术提供商的拥护和推崇,使得其迅速成为Web服务编排领域事实上的标准,从而使得无论是工业界还是学术界都对其产生了浓厚的兴趣,并围绕BPEL展开了多方面的理论与应用研究。
本文首先对BPEL的基本概念进行介绍,之后根据近年来各研究机构和个人对BPEL所作的研究将其分为如下几个研究方面:介绍Web服务支撑技术;工作流相关模式;形式化建模与验证;相关标准间的比较、转换与集成;面向服务的计算2;网格计算;自治计算进行分别介绍,最后对BPEL的研究进行总结与展望。
2BPEL概述2.1BPEL发展历史BPEL的前身是Microsoft的XLANG[1]和IBM的WSFL[2]。
“尽信书,则不如无书”—逆向思维与语言科学研究张国宪中国社会科学院语言所“尽信书,则不如无书”出自于古人孟子之口,告诫读书人要善于独立思考和分析,不要完全相信书本,更不能盲目地迷信书本。
相信书本是正向思维,怀疑书本是逆向思维。
逆向是与正向比较而言的,正向是指常规的、常识的、公认的或习惯的想法与做法。
逆向思维则恰恰相反,是对传统、惯例、常识的反叛,是对常规的挑战,所以逆向思维也叫求异思维,我们要强调的是,敢于“反其道而思之”,克服思维定势,破除由经验和习惯造成的僵化的认识模式,让思维向对立面的方向发展,从问题的反面深入地进行探索,创立新观点,这是每个科学研究者基本的、必备的思维方式。
“One who believes all of a book would be better off without books” comesfrom Mencius. He suggests that it is wise for intellectuals to have independentideas and analysis. Do not believe all of a book and never have blind faith in abook. Believing a book is normal thinkin g; doubting a book is “reversethinking”. Normal thinking refers to thinking which is done out of commonsense or habit. Reserve thinking, however, challenges tradition and routine.Every researcher should master reverse thinking for the reason that itovercomes the current paradigm, it breaks rigid cognitive models which arecaused by experience and habit, it helps to develop ideas in new ways, itanalyzes a situation from the other side of the coin, and most importantly, itcreates new ideas.1. 题解“尽信书,则不如无书”出自《孟子·尽心章句下》,意思是说完全相信书,那还不如没有书。
常用海运缩写————————————————————————————————作者: ————————————————————————————————日期:常用海运缩略语A始终保持浮泊AA Alwaysafloata.a.r.Against all保一切险risksACAccountc往来账户urrentA/CFor acco为…代销untof接受Acc.Acceptance;accepted根据ACCDGAccordingtoACCTAccount账,因为ACOL After装货结束后completionof ladingA.d.a/dAfterdate在指定日期后回扣佣金ADDCOM Address commissionAdd-on…Tariff费率(美国对非基本港口附加费)(alsoproportionalrate orarbitrary[in USA])从价Ad val. (a/v)Ad valorem(accordingtovalue)ADNE内河危险品国际运输欧洲规则uropean ProvisionsconcerningtheInternationalCarriageofDangerous Goodson Inland WaterwayADP Automate自动数据处理ddata processing公路危险货物运输欧洲协定ADR EuropeanAgreement concerningtheInternationalCarriage ofDangerousGoods byRoadADV Advise告知AETR European从事国际公路运输车辆从业人员工作的欧洲协定Agreement concerning the workofcrews ofvehiclesengagedininternationalroad transportAFRAAverage平均运费指数freight rate assessmentAFLWS As follows如下Agcy Agency代理公司Agt Agent代理人AGWAll goes一切顺利weighta.g.w.t.Actual gro实际毛重(总重)ss weightAHPS Arrival到达港口引航站harbor pilotstationAIMS American美国商船航运学会Instituteof Merchant ShippingAMT Air Mail T航空信汇ransferANCHAnchorage锚地ANS Answer回答…..账上A.O.Account ofAOHAfter off工作时间外icehoursA/or And/or及/或者账目付讫A/P Accountpaid附加保险费AP Additionalpremiumapprox.Approx近似,约计imatelyAPS Arrival pil到达引航站otstationA/Rall risks (in一切险surance)Arr arrival 到达(到港)Arrd.arrived已到达A/safter sight见票后付款(汇票)A/S alongsid在旁(船边)e尽快ASAP as soon aspossibleASBA Associa(美国)船舶经纪人和代理人协会tionofShipbrokers and Agents,Inc.ASCIIAmerican美国信息交换标准代码Standard Code forInformationInterchangeASPSArrival Sea到达海区引航站Pilot StationASS associate合伙人ATAactual tim实际到达时间eofarrivalATD actual time实际离开时间of departure包括任何时间,白天,夜间,星期天和节假日ATDNSHINC Any time,day,night,Sundays and holidaysincludedATP Agreeme易腐食品国际运输公约nt fortheInternationalCarriage of Perishable FoodstuffsATS All timesa所有节省时间vedATSBEAlltime s所有两港节省时间aved both endsATSDO All time所有卸货港节省时间daved dischargingonlyATSLO Alltime所有装货港节省时间savedlading onlyAttyattorney代理人,律师authauthorized授权的,许可的AuxAuxiliary辅助的,附件avaverage平均AWBAir Waybil空运单lAWTSAll working所有节省的工作时间timesavedAWTSBE All working 所有两港节省的工作时间timesaved bothendsAWTSDO Allw所有卸货港节省的工作时间orking timesaved discharging onlyAWTSLO All working所有装货港节省的工作时间time savedloading onlyBB.A.C.bunker a燃油附加费djustmentchargeB.A.F.bunker ad燃油附加费justmentfactorBags/BULKpart in ba袋装部分,散装部分gs,partinbulkBB Ballastb空航奖金(空放补贴)onusBBBBefore brea卸货前king bulkB.C.bulk cargo散货B/Dbank(er’银行汇票s) draft包括首尾两日b.d.i.both dates(days) inclusivebdth.breadth宽度Bdy.Boundary边界BENDSBoth ends两端(装货和卸货港)B/GBonded go保税货物odsBIMCOBaltic In波罗的海国际航运工会ternationalMaritime ConferenceBIZ Business 业务Bkge brokerage经纪费,佣金B/L bill oflo提单adingBlk.Bulk散(装)货Brl.Boiler锅炉BLTBuilt 建造BODBunkerof交燃油Delivery接燃油BOR Bunker ofRedeliveryBRLBarrel 桶(英,美容积单位)B/S Bill of sal卖契eBBSBasis基础BT Berth terms泊位条款BV Bureau法国船级社Veritas(FrenchShip ClassificationSociety)B.W.bonded wa保税仓库rehouseBxs.boxes盒,箱CCACcurrency货币(贬值)附加费adjustmentchargeCAConf Cargo货物代理会议(国际航空运输协会)Agency Conferenc e(IATA)CADcash aga凭单据付款inst documentsCAFCurrency货币贬值附加费adjustment factorCAP Capacity容量,能力CAPTCaptain船长CAScurrency货币贬值附加费adjustmentsurchargeCASS Cargo货物结算系统(国际航空运输协会)AccountsSettlementSystem(IATA)C.B.container集装箱基地baseCBF Cubic Feet立方英尺c.&d.collecti收运费交货on and deliveryc.b.d.cashbe付款后交货fore deliverycbmcubic metre立方米cccharges货到付款collectCCL customs cl结关earanceCCSconsoli集货(集装箱拼装货服务)dated cargo(contaiver)serviceccxcollect货到付款C/D customs海关申报单declarationCEM European c货车时刻表欧洲会议onference on goods trainTimetablesCEVNI European c欧洲内河航道代码ode forInlandWaterwaysCFMConfirm确认CFRCost and成本加运费(国际贸易术语解释通则)freight (Incoterms)CFScontainer集装箱货运站freight stationCGO Cargo货物C.H.carriers集装箱承运人接运haulageC.H.C.cargoha货物操作(装卸)费ndlingchargesCh.fwd.charges运费等货到后由收货人自付forwardCHOPT Chartere租方选择r’s optionCHTRSCharterers租船人c.i.a.cash in adv预支现金anceCIF cost,i到岸价(成本,保险费,运费)nsurance and freight(Incoterms)c.i.f.& e.cost,insu货价,保险费,运费,货币兑换rance,freight and exchange货价,保险费,运费,利息兑换c.i.f.i. & e.cost,insurance,,freight,interestandexchangec.i.f. &i.cost,insura货价,保险费,运费,利息nce,freight,and interestc.i.f.& c.cost,insur货价,保险费,运费,佣金ance,freight andcommissionc.i.f.c.& e.cost,货价,保险费,运费,佣金,兑换insurance,freight, commission and exchangec.i.f.c. &i.cost, insur货价,保险费,运费,佣金,利息ance,freight, commissionand interestc.i.f.i. &c.cost,in货价,保险费,运费,利息,佣金surance, freight,interestand commissionc.i.f.L.t.cost,ins货价,保险费,运费,伦敦条款uranceand freight,Londontermsc.i.f.w.cost, ins货价,保险费,运费及战险费urance andfreight plus war riskCIM Internatio铁路货物运输国际公约nal Conventiononthe TransportofGoods byRailwayCIP carriage运费及保险费支付到……andinsurancepaid toCIV Internati铁路乘客及行李运输国际公约onalConventionon theCarriageofPassenger and Luggageby RailwayCKD Complete全部拆散的(非装配的)ly knockeddown (unassembled)Cl.Classi分类,定级ficationCLC Civil国际由于损害民事责任公约Liability Conventionclean B/L clean Bill清洁提单of ladingcm centimeter公分(厘米)s(s)cm3Cubic cent立方厘米imeter(s)CMRConvention国际公路货运合同公约on the ContractfortheInternationalC/N Consignm发货通知书entnotecneeConsignee收货人cnmt/consgtConsign托付mentcnorConsignor发货人,委托人C/O Certifi产地证明书cateoforiginCOAContract包运合同of AffreightmentC.O.D.Cash on货到付款deliveryC.O.F.C.Conta平板车装运集装箱iner-on–Flat-Car (railflatcar)海上货物运输法COGSACarriageofGoodsby Sea ActCOM Commissi佣金onCONSCon消耗sumptionCVGPCustoms va海关价值,每磅毛重lue pergross poundCWECleared w海关免验结关ithout examination担,英=112磅美=100磅cwt HundredweightCWOCash with定货时付款/定购即付orderCYContainer集装箱堆场yardcy Currency货币D承兑交单D/AdocumentsagainstacceptanceDAF Delivered边境交货(国际贸易术语解释通则)at frontier (Incoterms)DAPdocument付款交单s againstpaymentD.A.S.delivered船边提货alongsideshipDBE(W)ATSDispatch支付两港所有节省(工作)的速遣费payableboth endson (working)all timesavedDCASDistribution费用分布分析系统Cost Anglysis SystemD.CLDetention 滞留条款ClauseD/CL.Deviation C绕航条款lause进干船坞D/D Dry dockingDDO Dispatch卸货港速遣discharging onlyDDP Delivered完税后交货(国际贸易术语解释通则)duty paid(Incoterms)DDU Delivered未完税交货(国际贸易术语解释通则)dutyunpaid (Incoterms)DELYDelivery 交付DEM Demurrag滞期(费)eDEPDeparture 离开,开航DES Delivered目的港船上交货(国际贸易术语解释通则)ex ship(Incoterms)DEQ Delivered目的港码头交货(完税货)(国际贸易术语解释通则)exquay(dutypaid)(Incoterms)DEVDeviation绕航DF DeadFre空舱费ightDFPDuty-free免税港portDHD Dispatch ha速遣费是滞期费的一半lfdemurrageDiaDiameter 直径DISCHDischarge 卸货DISPORT Discharge卸货港portDir.Direct直接的DLOSP Dropping 最后出港海区引航员下船last outward seapilotDm3Cubic d立方分米ecimeter(s)DO Diesel oil柴油CIM电子单证DOCIMELDocumentCIM Electronique(ElectronicCIMdocument)D/ODelivery提货单order出港引航员下船DOP DroppingoutwardpilotDOSP Dropping出港海区引航员下船outwardseapilotD/P Document付款交单sagainstpaymentDPP Dirty不洁石油产品PetroleumproductsDSTN Destination 目的地(港)DT Deep tank深舱DTG Distance t还要行使的距离ogoDTLSDetails细节DWCC Dead weight货物载重量cargo capacityDWTDead weigh载重吨ttonEE.& O.E.Errors and错误和遗漏不在此限omissionsexceptedDEIElectronic d电子数据交换ata interchange行政管理,商业和运输业的电子数据交换EDIFACTElectronic Data Interchangefor Administration,Commerce and Transport电子数据处理EDP Electronicdata processingE.g.Forexamp例如leEIR Equipment设备交接单(集装箱)interchange receipt(containers)Encl.Enclosure附件ERLOAD Expected预计准备装货ready to loadETAExpected ti预计到达时间me ofarrivalETB Expecte d预计靠泊时间time ofberthETCD Expected预计卸货完成时间time ofcompletionof dischargeETD Expected预计离开时间time ofdepartureETRExpected预计还船时间time of redeliveryETS Expected预计开航时间timeofsailingExcl.Excluding扣除EXW Ex works工厂交货(Incoterms)FFAAfree of all一切海损均不赔偿average一船舶能接受货交付的尽快速度FAC Asfastas thevessel can receive ordeliverf.a.c.fast ascan尽可能快的装卸(loadingordischarge)F.A.C.forwarding货运代理佣金agent’s commissionFACCOP Asfast as根据港口习惯以船舶尽可能快的装/卸货shipcanload/dischargeaccordingtocustom of port均一费率FAKfreight allkinds便利贸易特别项目(联合国贸发会议)FALPRO specialProgrammeon TradeFacilitation(UNCTAD)f.a.q.fair averag统货,大路货e qualityFASfree船边交货(国际贸易术语解释通则)alongsideship (Incoterms)FBL negotiableFIATA可转让多式联运提单FIATA Multimodal TrofLadingFCAFree carrie货交承运人(国际贸易术语解释通则)r(Incoterms)FCCfirstclas一流的租船人s charterersFCLfull con整箱货tainer loadFco.Franco;fre运费准免的eFDFreight运费和滞期费and demurrageFDNDFreight,di运费,速遣和滞期保护spatch anddemurragedefenceFFIFIATA ForFIATA运送指示wardingInstructionsf.g.a.freeof g共损不保eneral averageFHEXFriday an星期五和节假日除外dholidaysexceptedFHINCFridays and包括星期五和节假日holidaysincludedf.i.freein船方不负担装货费船方不负担装货费和理舱费f.i.a.s.free in andstowedFILOFree in Liner租船人承担装货费,船东承担卸货费outFILTD Free in Line租船人承担装货费,船东承担卸货费rtermsdischargeFIOFreeinan租船人承担装/卸货费doutFIOS Free in and 租船人承担装货、卸货和堆装费stowedFIOSTFree in and租船人承担装货、卸货、堆装和平舱费out and stowedandtrimmedFIOTFree inan租船人承担装货、卸货和平舱费doutand trimmedFIATA货运代理收货凭证FIATA FCR ForwardersCertificate of ReceiptFIATA货运代理运输凭证FIATAFCT ForwardersCertificateof TransportFIATASDT Shipper’sDFIATA托运人危险货物运输声明eclarationfortheTransporto f DangerousGoodsFIATA托运人联运重量证明书FIATA SIC ShippersIntermodalWeightCertificationF.I.B.Free into驳船交货barge运费,保险费f.i.c.freight, insurance, carriagef.i.h.free in港内交货harbourf.i.o.freein船方不负担装卸费用andoutf.i.o.s.freein船方不负担装卸费和理舱费and outstowedfiravv firstavail第一艘可用船舶able vesselFIS freight, insu运费、保险费和装船费ranceandshipping chargesf.i.w.free intow船方不负责装车费agonFLT forklift truc叉升式堆装机kFLWG Following 下达,下列FMFrom从,自FOFuelo燃料油/租船人承担卸货费,船东承担装货费il/Freeoutf.o.free out船方不负担卸货费FOBFree on船上交货,离岸价格boardF.O.C.flagsof方便旗船convenienceF.O.D.fre e of损害不赔damageFONASBA The Feder全国船舶经纪人和代理人协会联盟ation of National Associations ofShipbrokers andAgentsf.o.w.first第一解冻日open waterF.P.A Free of平安险particularaverageFPADfreight p到港支付运费ayable atdestinationFR Flat rack (co平底集装箱ntainer)FRT Freight运费Frt.fwd.freight f运费由提货人支付orwardFrt.ppd.freightp运费预付repaidFrt.ton freight ton运费吨Ft foot(feet)英尺FWFresh淡水waterFIATA不可转让的多式联运运单FWB non-negotiableFIATAMultimodaltransportWaybill满载重量和容量(集装箱)FWCfull loadedweight&capacityFWD Forward向前fwdr.forwarder货物代理人FIATA仓库收据FWRFIATAWarehouse Receipt供你指导FYGForyourguidanceFYIFor your给你提供信息informationFYR For your ref供你参考erenceGG.A.general a共同海损verageG.A.A.General共同海损担保AverageAgreementG.A.C.general共同海损摊款averag econtributionG.B.L.Government政府海运提单Bill ofLading杂货G.C.general cargoG.C.R.general ca杂货运价rgo ratesGDP gross domesticproduct国内生产总值Gds goods货物GFA GeneralFreightAgent货运总代理人GLESS Gearless无装卸设备GMT Greenwich Meantime格林威标准时间GNPGrossnationalproduct国民生产总值GRD Geard有装卸设备GRTGrossregisteredtonnage总登记吨Gr wtGross weight总重,毛重GSAGeneral SalesAgent销售总代理GST Goodsandservices tax货物和服务税Guatemala City Protocol(1971)Protocol toamend the Conventionfor the Unification of certain Rules relatingtoInternational CarriagebyAir signedat Warsaw on12October 19关于修改经1955年9月28日海牙议定书修改的1929年10月12日在华沙签署的《统一国际航空运输某些规则的公约》议定书(尚未生效)又称《1971年危地马拉议定书》29 as amended BytheProtocolat the Hague on28 September1955(not inforce)Guadalajara Supplementary Convention(1961)ConventionSupplementary to theWarsawConventionfor theUnification ofcertainRulesrelating toInternational carriageby Airperformedbya person otherthan thecontractingcarrier关于非由订约承运人履约的统一国际航空运输某些规则的华沙公约的补充公约,又称《1961年Guadalajara补充公约》HHagueProt.(1955)Protocolto amend the Convention forthe UnificationofcertainRulesrelating tothe International Carriage byAir signed a 关于修订1929年10月12日在华沙签订的《统一国际航空运输某些规则的公约》议定书又称《1955年海牙议定书》tWarsawin12 October1929Hague RulesInternationalConvention of theUnific action ofcertain Rulesrelating toBill ofLading (1924)《统一关于提单若干法律规定的国际公约》又称《海牙规则》Hague/Visby RulesProtocolto amendtheInternationalConventionfor the Unificationo f certainRules of Lawrelatingto BillsofLading(Brussels1968)《关于修订统一提单若干法律规定的国际公约议定书》(布鲁塞尔1968)又称《海牙-威斯比规则》Hamburg RulesUnited Nations Conventionon theCarriage ofGoods bySea(1978)《1978年联合国海上货物运输公约》《汉堡规则》HA/HOHatches/Holds舱口/舱HAWBHouse AirWaybill航空分运单HAZCHEM HazardousChemicals Code化学危险品代码HC HatchCoaming舱口围Hdlg Handling装卸欧洲铁路信息交换系统管理HERMESHandling European Railway MessageExchange SystemHFO Heavyfue重燃料油loilHgt Height高度H/lift Heavyli重型起动(设备)ft总局,总部H.Q.HeadquartersHRS Hours 小时HS Harmoni协调机制zed System(CCCConvention)HSD High Speed高速柴油disselHSSHeavygra重粮,黄豆,高粱in, soyas,sorghumH/STMG Harbor Ste港内航行amingIAC Including包括回扣佣金addresscommissioni.a.w.In accordan按照,根据ce withI.C.C.Institute C(伦敦保险协会货物保险条款)简称协会货物险条款argoClausesICD Inlandcle集装箱内陆验关堆场arance depotIFO Interme1000秒燃油diate fueloilIGSInertga惰性气体系统s systemIMDG Internati国际海运危险品符号onal MaritimeDangerousGoodsCode立即IMMImmediatelyIMO Internat国际海事组织ional Maritime OrganizationIn.Inch(es)英寸Incl.Including包括……..在内INCOTERMSStandard国际贸易术语解释通则(巴黎国际商会发出通知)conditionsfor saleand deliveryof goods (issuedby ICC,Paris)Info Information通知INMARSATInternation国际海事卫星组织国际公约al Conventionon theInternational MaritimeSatellite OrganizationInsInsurance 保险国际独立油轮船东协会INTERTANKO InternationalAssociationofIndependent TankerOwnersINTRMIntermediate中途点pointInvInvoice发票I.P.A.Including p包括单独海损articularaverageISDN Integrated ServicesDigitalNetwork综合服务数字网ITF International Transport Workers Federation国际海运劳工联盟IWL InstituteWarranty Limits协会保证区域限制Kkg(s)kilogram(s)千克km kilometer公里km.p.h kilometersper hour每小时公里km2square kilometer平方公里kn knot(s)海里,节(航速单位) kWkilowatt千瓦KWhkilowatt-hour千瓦/小时KWSknots节/海里KyotoConventionInternational Conventionon theSimplification andHarmonization ofCustomsProcedures (1973)(CCC Convention)1973年简化和协调海关程序国际公约(CCC公约)又称《西京公约》LL/A Lloyd’s agent劳埃德保险公司代理人LASH lighter载驳货船/子母船aboard shipLat.,latLatitude纬度LAYCANLaydays船舶受载期和解约日andcanceling datelb(s)pound(s)磅L/C Letter of cr信用证edit装货和卸货L/D Loading anddischargingl.&d.loss and d灭失和损坏amagel.& u.loading a装和卸ndunloadingLCL less thanc拼装箱货,拼装车货(零担货)ontainerload(lessthan carload)LDGLoading 装货LELlower ex最低爆炸点plosive limit最低燃点LFLlower flammablelimitlgt.Long ton;长吨long tons船东承担装货费,租船人承担卸货费LIFOLiner infree outliq.liquid液态Lkg/Bkg leakage & br漏损和破损eakage液化天然气LNGLiquefiednatural gasL.O.A.length o全长ver allLO/LOlift on .L吊上吊下方式iftoffloc.local;locatio当地的,定位n保函LOI Letter o fIndemnityLong., long longitude经度,经线LPG Liquefie液化石油化学气dpetrochemicalgasLSD landin卸货,存储和交货费g,storageand delivery chargesL.T.local time当地时间L/T linerter班轮条款msLTBENDSLiner船东承担装货和卸货费terms bothendsltr.lighter驳船LUBOIL Lubricat润滑油ion oillump lump sum包干费MM Minimum(r起码(费率等级)ate classification)mmetre (s)公尺m3cubicme立方米tre(s)MACH modular定型自动化集装箱搬运automatedcontainerhandlingMARPOLMaritime A有关油污责任的海运协定greement Regarding OilPollutionof Liabilitymat material资料MAWB Master Air 航空主运单WaybillMDOMediumdiesel oil混合柴油Mdse Merchandise商品MED Mediterranean地中海MEGC’sMultipleElement Gas Containers多元化气集装箱MFN Most FavouredNation最惠国M.H.MerchantsHaulage(集装箱)货方接运MIAMarine InsuranceAct海上保险法M/Rmate’sreceipt大副收据M+R Maintenance and repair (center)维修保养(中心)Mi mile(s)英里MOLCO More or lessowner’s option船东选择或多或少Montreal ConventionMontreal Prot.No.4 (1975)Conventionfor the Unificationofcertain Rulesrelating toInternationalCarriage by Air(1999)Protocol toamendtheConvention for theUnificat统一国际航空运输某些规则的公约(1999)。
《国防科技大学学报》参考文献著录规范与实例1)参考文献表只列出在正文中被引用过的、新的、重要的、正式发表的文献资料,数量一般不少于15 条,注意列出对《国防科技大学学报》的引用。
采用顺序编码制组织,各篇文献按正文中依次标注的序号列出。
2)英文参考文献只用英文著录。
作者的名字一律“姓”全拼在前,“名”缩写成首字母在后,缩写“名”后省略缩写点。
题名、书名等只是句首字母大写,期刊、专著、论文集等文献名中各实词首字母大写。
3)中文参考文献同时用中、英文著录,在英文后加上(in Chinese)。
注意:在对照的英文中,作者姓名录入格式和篇首页作者姓名格式一致,其余同英文参考文献著录格式一致。
4)每篇文献最多列出3位作者,多于3位时,中文后加“,等”字,英文后加“,et al”,姓名之间用“,”分隔。
5)会议论文一律以论文集的形式著录,特别注意需注明论文集的出版地、出版者和出版年,一般不再写会议的举办地点和时间。
论文集没有正式出版,就顺序列出会议论文集名称、举办地点和时间。
6)参考文献中以下常用缩写尽可能改为相应的全拼Proc—Proceedings,Symp—Symposium,Trans—Transactions,Conf—Conference7)每条文献中各必选项必须齐全,并注意各项的排列顺序和每项之后标点的用法。
8)若引用的参考文献尚在在印刷中,在条目最后加上“:印刷中”或“: in Press”。
9)按文献的出版形式划分,可将文献分为普通图书、期刊、报纸、会议文集、学位论文、科技报告、技术标准、专利、电子资源等。
按文献的存储载体不同,又可将文献分为纸质、磁带、磁盘、光盘和联机网络等,没有特别声明的文献均属于纸质文献。
各标志代码如下:10)电子文献不仅要著录文献类型,而且必须注明文献载体类型;联机文献必须注明引用日期( [yyyy-mm-dd] )、获取和访问路径。
注:在著录格式中:1. { }中的部分是可选项;蓝色部分对于联机文献是必选项。
国际会议发言稿国际会议发言稿第一篇:国际学术会议发言稿1. prol oguetha nk you, mr. ch airman, for yo ur grac iousin troduct ion. iam hono red tohave th e chanc e to ad dress y ou on t his spe cial oc casion.the to pic ofmy pape r is “t ransact ion cos t and f armers’choice of agr icultur al prod ucts se lling”. the ou tline o f my ta lk as f ollows. the fi rst par t i wan t to in troduce the ba ckgroun d of th is rese arch. t he seco nd part sugges ts a si mple ho usehold choice model.the th ird par t cover s the d ata use d in th is rese arch. a nd then, we in troduce the em pirical result s. fina lly, asimpleconclus ion isgiven.2. in troduct ionwell, let’s move o n the f irst pa rt of t his top ic .the motiva tion of this w ork lik e this. instit utional econom ics pos its tha t agent s makin g decis ions on differ ent typ es of t ransact ions do so ina costl y way .for exa mple ,farmers decidi ng sell a part icularcrop to whom b ase the ir deci sions n ot only on the pricethey ex pect to receiv e in ea ch mark et choi ce butalso on additi onal co sts rel ated totransa cting i n these market s.i wan t to us e a pic ture toillust rate it. for e xample, givensome ma rket ch annels, farmer s’ choi ces can be reg arded a s equil ibriumbetween the su rplus a nd theadditio nal cos ts that relate d totr ansacti ng .esp ecially in dev eloping countr ies, hi gh-valu e cropproduce rs full y parti cipatein themarketand thetransa ction c ost has been t he hard constr aint to farmer s. furt hermore, farme rs’ mar ket cho ices ca n be ta ken asa choic e dilem ma of t ransact ion cos t and p roducti on surp lus.co nsequen tly, th e scien tific q uestion of thi s resea rch ishow tra nsactio n costaffects plante rs’ cho ices. 3. met hodolog ylet’smove to the th eoretic al mode l of ou r resea rch. co nsidera house hold mo del inone rot ation.in stag e1 , f amer ηneeds t o alloc ate the inputfactors.thisprocess can qb e set i nto a f unction like t his q?? q( p, w , z ? ) ,qη mean s the o utput f armersdecideqto pro duce .p implie s the o utput p rice wimplies inputprice a nd.z: ?is fix ed inpu t. once produc e whatand pro duce ho w manyare dec ided, n extque stion t o be co nsidere d is ho w muchproduct s to be transa ctedin market. herewe usethree c c()cηm eans ho w funct ions to descri be this questi on. the firstequatio n, c ??p ,z?much a gricult ural pr oductsused by famers themse lves. pimplie s the p rice th e cagri cultura l produ ct,z ?s uggests the fl uctuati on of cη. thesecondequatio n q ? ?q ? ?c?, qηm eans th e amoun t of ag ricultu ral pro ducts t ransact ed inq?n?marke t. thethird e quation i ? q ?implies the am ount ex changed in nth time.i n stage 3, far mers wi ll deci de to s ell the produc tsto w hom. ch anel j’s marke t price isbdec ided by an exo genesispriceand far mers’ n egotiat ingpow er.pij?p*j?b(q i,zi)be sides t his, we use amatrixto show the ne t profi t of ch anel jx ik? ik, ? ??ik ?and then f armers’choice can be expres sed ina typic al choi cemode lexp(xi j?)pr(j i?j|xik)?1 exp(xij?)? k?1bas ed on t hechoi ce mode l, anot her imp ortantconcept is fam ers’ch annel c hoice .here, w e set f ive typ es .the y rankby themarketbarrier s. acco rdingly, we se t a gro up disc rete nu mber to expres s them. y: dep endentvariabl e y=5,m eans fa rmerch oosebro kers. y=1, far mers se ll prod ucts to consum ersdir ectly.4. da ta andestimat ion pro cedures here, w e illus trate t he datadistri butionwith th is map. accord ing tothe agr icultur al regi onaliza tion fr om depa rtmentof agri culture, the a pplesp ecializ ation a reas in chinacontain two pa rts: bo sea ar ea andloess p lateau. bo sea area i n red c olor, c ontains hebei, shando ng andliaonin g 3 pro vinces.and lo ess pla teau in greencolor,contain s shanx i, hena n, shaa nxi and gansu4provi nces. f irstly, we use pps me thod to get th e first stagesamplin g unit14 coun ties in7 prov inces.then us e rando m sampl e metho d to ge t villa ge andhouseho ld. the y are o ur samp ledist ributio n.5. empiri cal res ults6. conc lusions第二篇:国际会议国际会议:201X 2n d inter nationa l confe rence o n envir onmentscience and en gineeri ng(ices e 201X)post by:201X-7-24 19:09:28 [只看该作者]201X 2nd in ternati onalco nferenc e on en vironme nt scie nce and engine eringma y. 26,201X ne ws! the icese201X pa pers ar e avail able in the vo l.8 ofipcbee. (click)mar.16, 201Xnews! t he conf erenceprogram of ice se 201X is ava ilablenow. (c lick)we le to t he offi cial we bsite o f the 201X 2nd intern ational confer ence on enviro nment s cienceand edduringapril 7-8 201X, in ba ngkok,thailan d. ices e 201X, aims t o bring togeth er rese archers, scien tists,ehe pra cticalchallen ges enc ountere d and t he solu tions a dopted.changeand sha re thei r exper iences,new id eas, an d resea rch res ults ab out all aspect s of en vironme nt scie the con ference will b e heldevery y ear tomake it an ide al plat form fo r peopl e to sh are vie ws andexpsens ors and relate d areas. icese201X w ill bepublish ed in o ne volu me of i pcbee,and all papers will b e y dig italli brary,and ind exed by thomso n isi (web ofknowled ge).cal l for p apers201X 2ndinterna tionalconfere nce onenviron mentsc ience a nd engi neering, icese 201X i s the p remierforum f or esea rch res ults in the fi elds of theore tical,experim ental,and app lied en vironme nt scie nce and engine ering.the con fereers, engin eers an d scien tists i n the d omain o f inter est fro m aroun d the w orld. t opics o f inter est for submis sionin clude,e nvironm ental s cienceand tec hnology environ mentaldynamic smeteor ologyhy drology geophys icsatmo sphericphysic sphysic al ocea nograph yglobal enviro nmental change and ec osystem s manag ementcl imate a nd clim atic ch angesgl obalwa rmingoz one lay er depl etionca rbon ca pture a ndstor agebiof uelsint egrated ecosys tems ma nagemen tsatell iteapp licatio ns in t he envi ronment environ mentalrestora tion an d ecolo gical e ngineer inghabi tat rec onstruc tionbio diversi ty cons ervatio ndefore station wetland slandsc ape deg radatio nand r estorat iongrou nd wate r remed iations oildec ontamin ationec o-techn ologybi o-engin eeringe nvironm ental s ustaina bilityr esource manage mentlif e cycleanalys isenvir onmenta l syste ms appr oachren ewablesources of ene rgy-ene rgy sav ingscle antech nologie ssustai nable c itieshe alth an dthe e nvironm entheal th rela ted org anismsh azardou ssubst ances a nd dete ction t echniqu esbiode gradati onof h azardou s subst ancesto xicityassessm ent andepidem iologic al stud iesqual ity gui delines, envir onmenta l regul ation a nd moni toringi ndoor a ir poll utionwa ter res ourcesand riv er basi n manag ementre gulator y pract ice, wa ter qua lity ob jective s stand ard set ting, w ater qu ality c lassifi cationpublicpartici patione conomic instru mentsmo delinganddec ision s upporttoolsin stituti onal de velopme nttrans boundar y coope rationm anageme nt andregulat ion ofpoint a nd diff use pol lutionm onitori ng andanalysi s of en vironme ntalco ntamina ntgroun d water manage mentwas tewater and sl udgetr eatment nutrien ts remo valsusp ended a nd fixe d filmbiologi cal pro cessesa naerobi c treat mentpro cess mo delling sludgetreatme nt andreusefa te of h azardou s subst ancesin dustria l waste water t reatmen tadvanc es in b iologic al, phy sical a ndchem ical pr ocesses on site and sm all sca le syst emsstor m-water manage mentair pollut ion and contro lemissi onsour cesatmo spheric modeli ng andnumeric alpred ictioni nteract ion bet ween po llutant scontro ltechn ologies air emi ssion t radings olid wa steman agement waste m inimiza tionopt imizati on of c ollecti onsyst emsrecy cling a nd reus ewastevaloriz ationte chnical aspect s of tr eatment and di sposalmethods (landf illing, therma ltreat ment et c) leac hate tr eatment legal,economi cand m anageri al aspe cts ofsolid w aste ma nagemen tmanage ment of hazard ous sol id wast ewatertreatme nt andreclama tiondis infecti on anddisinfe ction b y-prod uctsman agement of wat er trea tment r esidual saesthe ticqua lity of drinki ng wate r (tast e, odor s)effec t of di stribut ion sys tems on potabl e water qualit yreuseof recl aimed w atersde xed bythomson isi (w eb of k nowledg e).impo rtant d ateadv anced t reatmen t of wa ter and second ary eff luents(membra nes, ad sorptio n, ionexchang e, oxid ation e tc) ice se 201X will b e publi shed in one vo lume of ipcbee, and a ll pape rs will beinc luded i n engin eeringp aper su bmissio n (full paper)beforedecembe r 10, 201X not ificati on of a cceptan ceon ja nuary 1, 201Xfinal p aper su bmissio nbefore januar y 15, 201Xaut hors' r egistra tionbef ore jan uary 15, 201Xicese 201Xcon ference datesa pril 7- 8, 201Xsubmi ssion m ethods:1. e lectron ic subm issionsystem; ( .pdf)if you can'tlogin t he subm issionsystem, please try to submit throug hmetho d2.2. e mail: i cese@cb ees.org ( .pdf and .d oc)第三篇:国际会议invitat ion let terdear profes sor li,on beha lf of t he peki ng univ ersityand the ieee p uter so ciety,i would be ver y pleas ed to i nvite y ou to a ttend a nd chai r a ses sion of thefo rthing201X in ternati onal co nferenc e on pa ralleldatapr ocessin g to be held i n beiji ng, fro m march 25 tomarch 28, 201X.you ar e an in ternati onallyacclaim ed scho lar and educat or. you r parti cipatio n willbe amon g the h ighligh ts of t heconf erence.we sinc erely h ope tha t you c ould ac cept ou rinvit ation.if youcan e,pleaselet usknow as soon a s possi ble, si nce wehave to prepar e the f inal pr ogram s oon. we arelo oking f orwardto your accept ance.si ncerely yours,wangya ngreply letter dear pr ofessor wang,m any tha nks foryour l etter d ated 15th febr uary, i nviting me toattendand cha ir a se ssion o f the f orthing 201X i nternat ional c onferen ceon p arallel data p rocessi ng to b e heldin beij ing, fr om marc h 25 to march28, 201X.muchto my r egret,i shall not be able t o honor the in vitatio n becau se i ha ve been suffer ing fro m adis ease si nce thi s summe r. i am firmly advise d thatit woul d be un wise to undert ake any distan t and l ong tra vel inthe nea r futur e.i fee l verysad tomiss th e oppor tunityof meet ing you and ma ny othe rs in t he fiel d of pu ter sci ence. i wish t he conf erencea plete succes s. sinc erely y ours,limingin quiry l etterde ar sir,thanksyou ver y muchfor acc eptingour inv itation. i amwriting to ask for in formati on abou t yourbeijing travel for th e sessi on. you will a rriverat marc h 24, a nd plea se letme know that y our arr ival ti me andspecifi cways, so tha t we ma y arran ge pers on to p ick you up. we are lo oking f orwardto your arriva l.since rely yo urs,wan g yang第四篇:国际会议中心国际会议中心■长隆酒店配有国际会展中心,这是华南地区最大的国际会展中心之一,独立的一整栋楼供您举办商务宴会、商务会议、主题婚宴等等。
Bibliography[1] A.M.Abo,Design for reliability of low-voltage,switched-capacity circuits.Ph.D.Thesis,University of California,Berkeley,1999[2] A.M.Abo,P.R.Gray,A1.5-V,10-bit,14.3-MS/s CMOS pipeline analog-to-digital converter.IEEE J.Solid-State Circuits34(5),599–606(1999)[3] B.K.Ahuja,An improved frequency compensation technique for CMOS operational ampli-fiers.IEEE J.Solid-State Circuits18(6),629–633(1983)[4]V.J.Arkesteijn,Analog front-ends for software-defined radio receivers.Ph.D.dissertation,University of Twente,2007[5]V.J.Arkesteijn,E.A.M.Klumperink,B.Nauta,Jitter requirements of the sampling clock insoftware radio receivers.IEEE Trans.Circuits Syst.(TCAS)II53(2),90–94(2006)[6] C.W.Barbour,Simplified PCM analog to digital converter using capacity charge transfer,inProc.of the Telemetering Conf.(1971),pp.4.1–4.11[7] A.Barna,D.I.Porat,Integrated Circuits in Digital Electronics(Wiley,New York,1973),pp.353–354[8]W.C.Black,D.A.Hodges,Time interleaved converter arrays.IEEE J.Solid-State Circuits15(6),1022–1029(1980)[9]M.Boulemnakher,E.Andre,J.Roux,F.Paillardet,A1.2V4.5mW10b100MS/s pipelinedADC in65nm CMOS,in ISSCC Dig.Tech.Papers(2008),pp.250–251[10]T.B.Cho,P.R.Gray,A10b,20Msample/s,35mW pipeline A/D converter.IEEE J.Solid-State Circuits30(3),166–172(1995)[11]R.H.Dennard,F.H.Gaensslen,H.N.Yu,V.L.Rideovt,E.Bassous,A.R.LeBlanc,Design ofion-implanted MOSFET’s with very small physical dimensions.IEEE J.Solid-State Circuits 256–268(1974)[12]S.Devarajan,L.Singer,D.Kelly,S.Decker,A.Kamath,P.Wilkins,A16b125MS/s385mW78.7dB SNR CMOS pipeline ADC,in ISSCC Dig.Tech.Papers(2009),pp.86–87 [13] A.G.F.Dingwall,Monolithic expandable6bit20MHz CMOS/SOS A/D converter.IEEE J.Solid-State Circuits14(6),926–932(1979)[14]M.El-Chammas,B.Murmann,General analysis on the impact of phase-skew in time-interleaved ADCs,in IEEE International Symposium on Circuits and Systems(ISCAS) (2008),pp.17–20[15]S.L.J.Gierkink,Control linearity and jitter of relaxation oscillators.Ph.D.dissertation,Uni-versity of Twente,1999[16]S.K.Gupta,M.A.Inerfield,J.Wang,A1-GS/s11-bit ADC with55-dB SNDR,250-MAWpower realized by a high bandwidth scalable time-interleaved architecture.IEEE J.Solid-State Circuits41(12),2650–2657(2006)[17]K.Hadidi,A.Khoei,A highly linear cascode-driver CMOS source-follower buffer,in IEEEIntl.Conf.on Electronics,Circuits and Systems(1996),pp.1243–1246S.M.Louwsma et al.,Time-interleaved Analog-to-Digital Converters,131 Analog Circuits and Signal Processing,DOI10.1007/978-90-481-9716-3,©Springer Science+Business Media B.V.2011132Bibliography [18] C.-C.Hsu,F.-C.Huang,C.-Y.Shih,C.C.Huang,Y.-H.Lin,C.-C.Lee,B.Razavi,An11b800MS/s time-interleaved ADC with digital background calibration,in ISSCC Dig.Tech.Papers(2007),pp.464–465[19] E.Iroaga,B.Murmann,L.Nathawad,A background correction technique for timing errors intime-interleaved analog-to-digital converters,in IEEE International Symposium on Circuits and Systems(ISCAS),vol.6(2005),pp.5557–5560[20] E.A.M.Klumperink,B.Nauta,Systematic comparison of HF CMOS transconductors.IEEETrans.Circuits Syst.II,Analog Digit.Signal Process.50(20),728–741(2003)[21]N.Kurosawa,H.Kobayashi,K.Maruyama,H.Sugawara,K.Kobayashi,Explicit analysisof channel mismatch effects in time-interleaved ADC systems.IEEE Trans.Circuits Syst.I, Fundam.Theory Appl.48(3),261–271(2001)[22] F.Kuttner,A1.2V10b20MSample/s non-binary successive approximation ADC in0.13µmCMOS,in ISSCC Dig.Tech.Papers(2002),pp.176–177[23]Y.Z.Lin,S.J.Chang,Y.T.Liu,C.C.Liu,G.Y.Huang,A5b800MS/s2mW asynchronousbinary-search ADC in65nm CMOS,in ISSCC Dig.Tech.Papers(2009),pp.80–81 [24]S.M.Louwsma,E.J.M.van Tuijl,M.Vertregt,B.Nauta,A1.6GS/s,16times interleavedtrack&hold with7.6ENOB in0.12µm CMOS,in Proc.ESSCIRC(2004),pp.343–346 [25]S.M.Louwsma,E.J.M.van Tuijl,M.Vertregt,B.Nauta,A1.35GS/s,10b,175mW time-interleaved AD converter in0.13µm CMOS,in Proceedings of the Symposium on Very Large Scale Integration(VLSI)Circuits(2007),pp.62–63[26]S.M.Louwsma,E.J.M.van Tuijl,M.Vertregt,B.Nauta,A time-interleaved track&holdin0.13µm CMOS sub-sampling a4GHz signal with43dB SNDR,in Proceedings of the Custom Integrated Circuits Conference(CICC)(2007),pp.329–332[27]S.M.Louwsma,A.J.M.van Tuijl,M.Vertregt,B.Nauta,A1.35GS/s,10b,175mW time-interleaved AD converter in0.13µm CMOS.IEEE J.Solid-State Circuits43(4),778–786 (2008)[28]J.McCreary,P.Gray,All-MOS charge redistribution analog-to-digital conversion techniques.IEEE J.Solid-State Circuits10(6),371–379(1975)[29] E.Mensink,E.A.M.Klumperink,B.Nauta,Distortion cancellation by polyphase multipathcircuits.IEEE Trans.Circuits Syst.I,Regul.Pap.52(9),1785–1794(2005)[30]G.E.Moore,Cramming more components onto integrated circuits,in Electronics(1965),pp.114–2117[31] B.Murmann,A/D converter trends:power dissipation,scaling and digitally assisted architec-tures,in Proceedings of the Custom Integrated Circuits Conference(CICC),(2008),pp.751–758[32] B.Murmann,ADC Performance Survey1997–2009.Available online:http://www./~murmann/adcsurvey.html(2009)[33]K.Nagaraj, D.A.Martin,M.Wolfe,R.Chattopadhyay,S.Pavan,J.Cancio,T.R.Viswanathan,A dual-mode700-Msamples/s6-bit200-Msamples/s7-bit A/D converter in a0.25-µm digital CMOS process.IEEE J.Solid-State Circuits35(12),1760–1768(2000) [34]Y.Nakagome,H.Tanaka,K.Takeuchi,E.Kume,Y.Watanabe,T.Kaga,Y.Kawamoto,F.Murai,R.Izawa,D.Hisamoto,T.Kisu,T.Nishida,E.Takeda,K.Itoh,Experimental1.5-V 64-Mb DRAM.IEEE J.Solid-State Circuits26(4),465–472(1991)[35] B.Nauta,Analog CMOS low power design considerations,in Low Power Workshop on ES-SCIRC Conference(1996)[36] B.Nikoli´c,V.G.Oklobdžija,V.Stojanovi´c,W.Jia,J.K.S.Chiu,M.M.T.Leung,Improvedsense-amplifier-basedflip-flop:design and measurements.IEEE J.Solid-State Circuits35(6), 876–884(2000)[37]H.Pan,M.Segame,M.Choi,J.Cao,A.A.Abidi,A3.3-V12-b50-MS/s A/D converter in0.6-µm CMOS with over80-dB SFDR.IEEE J.Solid-State Circuits35,1769–1780(2000) [38]M.J.M.Pelgrom,A.C.J.Duinmaijer,A.P.G.Welbers,Matching properties of MOS transis-tors.IEEE J.Solid-State Circuits24(5),1433–1439(1989)Bibliography133 [39]K.Poulton,J.J.Corcoran,T.Hornak,A1-GHz6-bit ADC System.IEEE J.Solid-State Cir-cuits22(6),962–970(1987)[40]K.Poulton,R.Neff,B.Setterberg,B.Wuppermann,T.Kopley,R.Jewett,J.Pernillo,C.Tan,A.Montijo,A20GS/s8b ADC with a1MB memory in0.18µm CMOS,in ISSCC Dig.Tech.Papers(2003),pp.318–496[41] B.Razavi,Rf Microelectronics(Prentice Hall,New York,1998)[42] D.Schinkel,E.Mensink,E.A.M.Klumperink,A.J.M.van Tuijl,B.Nauta,A double-taillatch-type voltage sense amplifier with18ps setup+hold time,in ISSCC Dig.Tech.Papers (2007),pp.314–315[43] D.Schinkel,E.Mensink,E.A.M.Klumperink,A.J.M.van Tuijl,B.Nauta,A low-offsetdouble-tail latch-type voltage sense amplifier,in Proceedings of the18th ProRisc Workshop (2007)[44]H.Schmidt,Analog-Digital Conversion(Van Nostrand-Reinholt,New York,1970)[45] A.J.Scholten,G.D.J.Smit,B.A.D.Vries,L.F.Tiemeijer,J.A.Croon,D.B.M.Klaassen,R.van Langevelde,X.Li,W.Wu,G.Gildenblat,The new CMC standard compact MOS model PSP:advantages for RF applications,in IEEE Radio Frequency Integrated Circuits Symposium(2008),pp.247–250[46]T.Sepke,P.Holloway,C.G.Sodini,H.S.Lee,Noise analysis for comparator-based circuits.IEEE Trans.Circuits Syst.I,Regul Pap.56,541–553(2009)[47]R.C.Taft,P.A.Francese,M.R.Tursi,O.Hidri,A.MacKenzie,T.Hoehn,P.Schmitz,H.Werker,A.Glenny,A1.8V1.0GS/s10b self-calibrating unified-folding-interpolatingADC with9.1ENOB at Nyquist frequency,in ISSCC Dig.Tech.Papers(2009),pp.78–79 [48]H.P.Tuinhout,G.Hoogzaad,M.Vertregt,R.L.J.Roovers,C.Erdmann,Design and character-ization of a high precision resistor ladder test structure,in Proceedings of the IEEE Interna-tional Conference on Microelectronic Test Structures(ICMTS),vol.15(2002),pp.223–228 [49]R.C.H.van de Beek,High-speed low-jitter frequency multiplication in CMOS.Ph.D.disser-tation,University of Twente,2004[50]R.J.van de Plassche,Integrated Analog-to-Digital and Digital-to-Analog Converters(Kluwer Academic,Dordrecht,1994)[51]R.J.van de Plassche,R.E.J.van der Grift,A high-speed7bit A/D converter.IEEE J.Solid-State Circuits14(6),938–943(1979)[52]G.van der Plas,B.Verbruggen,A150MS/s133µW7b ADC in90nm digital CMOS using acomparator-based asynchronous binary-search sub-ADC,in ISSCC Dig.Tech.Papers(2008), pp.242–243[53]H.van der Ploeg,Calibration techniques in two-step a/d converters.Ph.D.dissertation,Uni-versity of Twente,2005[54]M.van Elzakker,A.J.M.van Tuijl,P.F.J.Geraedts,D.Schinkel,E.A.M.Klumperink,B.Nauta,A1.9µW4.4fJ/conversion-step10b1MS/s charge-redistribution ADC,in ISSCCDig.Tech.Papers(2008),pp.245–245[55] A.J.M.van Tuijl,personal communication[56] A.Verma,B.Razavi,A10b500MHz55mW CMOS ADC,in ISSCC Dig.Tech.Papers,(2009),pp.84–85[57]M.Vertregt,The analog challenge of nanometer CMOS,in International Electron DevicesMeeting(IEDM)(2006),pp.1–8[58]M.Vertregt,H.P.Tuinhout,personal communication[59]Video-transcript,Excerpts from a conversation with Gordon Moore:Moore’s law.ftp:///museum/Moores_Law/Video-Transcripts/Excepts_A_Conversation _with_Gordon_Moore.pdf(2005)[60]R.H.Walden,Analog-to-digital converter survey and analysis.IEEE J.Sel.Areas Commun.17(4),539–550(1999)[61] B.Wicht,T.Nirschl,D.Schmitt-Landsiedel,Yield and speed optimization of a latch-typevoltage sense amplifier.IEEE J.Solid-State Circuits39(7),1148–1158(2004)134Bibliography [62]Wikipedia,Orthogonal frequency-division multiplexing./wiki/Orthogonal_frequency-division_multiplexing(2009)[63]K.L.J.Wong,C.K.K.Yang,Offset compensation in comparators with minimum input-referred supply noise.IEEE J.Solid-State Circuits39(5),837–840(2004)[64]W.Yang,D.Kelly,L.Mehr,M.T.Sayuk,L.Singer,A3-V340-mW14-b75-Msample/sCMOS ADC with85-dB SFDR at Nyquist input.IEEE J.Solid-State Circuits36(12),1931–1936(2001)Index3D EM-field simulation,16AAD-Convertercounting,40flash,3,39,93folding,39pipeline,28,29,31,40,45,48,59,63–68, 93SA-ADC,40–57,91,94–108slope,40two-step,40amplifier,40,59,64,92,93,122interstage,108–111architectureTrack&Holdwith frontend sampler,17–20without frontend sampler,13–17Bbandwidthinput,16,19,30body effect,23bootstrapping,78–85bottom-plate sampling,6,28,29bufferbandwidth requirement,26distortion,23implementation,90input,14open-loop,22source follower,23,24,26,38,90,91Ccalibration,32–35,58,72,85,113,115,118 background,33bandwidth,12,22,35foreground,33gain,34,114offset,34,57,63,114,115timing,34,121,123 capacitancebuffer,input,25input,13,15–17interconnect,9capacitive load,26channel-charge injection,80,84,85 charge redistribution,27,83,110,111 clock feed-through,84,85clock generation,68,72,73,75,88,95 comparator,54–57,59,97,98Ddecoder,105digital control,40,41,94,99Eerrorgain,5offset,5timing,5Ffeedback,22Hhold-mode,5Jjitter,34–37,73,78,85,120,121,123S.M.Louwsma et al.,Time-interleaved Analog-to-Digital Converters,Analog Circuits and Signal Processing,DOI10.1007/978-90-481-9716-3,©Springer Science+Business Media B.V.2011135136IndexLladder connections,106layout,15,16,86,116look-ahead logic,53,99,101,103Mmatchingcapacitor,10Miller effect,25,109mismatchbandwidth,9–12between channels,6gain,6,7offset,6timing,6Nnoiseamplifier,64kT/C,16,58–61,64,85,111variance,59,64–66non-interleaved,5,6,22,26,30,39,68Ooffsetchannel,6comparator,57opamp,28,29,31,57,58,63,67,91,93,104, 108,109Pphase-differences,8Rreliability,79,82,83reset switch,15resistanceinterconnect,9switch,9,10,19,51,79Ssettling,14settling time,19,28,41–44,48,49single-sided overrange technique,46,47,49, 94,99,101spectrum,6spurious tones,6switchto avoid distortion,27switch-driver,85Ttechnology,10,14,16,19,22,32timing-misalignment,8,17Track and Holdbuffer,22–28track-mode,5track-timereduction,14,18track-time reduction,29 transconductance amplifier,59 transmission lines,13。
P H.D. T HESIS – M ATS D ANIELSON, KTH, S TOCKHOLM, S WEDENReferences[A53]Allais, M.: “The Foundations of a Positive Theory of Choiceinvolving Risk and a Criticism of the Postulates and Axioms ofthe American School” in Expected Utility Hypothesis and the AllaisParadox, D. Reidel Publishing Company, 1979 (Originally inFrench 1953).[B90]Bana e Costa, C.A.: Readings in Multiple Criteria Decision Aid,Springer-Verlag, 1990.[BCG87]Barzilai, J., Cook, W., and Golany, B.: “Consistent Weights for Judgement Matrices of the Relative Importance of Alternatives”in Operations Research Letters, vol.6, pp.131–134, 1987.[BZ70]Bellman, R., and Zadeh, L.A.: “Decision Making in a FuzzyEnvironment” in Management Science, vol.17, pp.B-144–B-164, 1970. [BG83]Belton, V., and Gear, A.E.: “On a Shortcoming of Saaty’s Method of Analytical Hierarchies” in OMEGA, vol.11, pp.227–230, 1983.[B54]Bernoulli, D.: “Exposition of a New Theory on the Measurement of Risk” in Econometrica, vol.22, pp.23–36, 1954 (Originally from 1748). [B91]Bertsekas, D.: Linear Network Optimization, MIT Press, 1991.[BC92]Bicchieri, C., and Chiara, M.L.D.: “Preface” in Knowledge, Belief, and Strategic Interaction, eds. Bicchieri and Chiara, pp.vii–xii,Cambridge University Press, 1992.[B95]Boman, M.: “Rational Decisions and Multi-Agent Systems” in Proc.Working Notes for the AAAI Fall Symposium on Rational Agency,ed. Fehling, MIT Technical Report, 1995.[BE94]Boman, M., and Ekenberg, L.: “Eliminating Paraconsistencies in 4-valued Co-operative Deductive Multidatabase Systems with ClassicalNegation” in Proceedings of Cooperating Knowledge Based Systems1994, pp.161–176, Keele University Press, 1994.[BE95]Boman, M., and Ekenberg, L.: “Decision Making Agents withRelatively Unbounded Rationality” in Proc. DIMAS’95,pp.I/28–I/35, 1995.[BG88]Bond, A.H., and Gasser, L.: Readings in Distributed ArtificialIntelligence, Morgan Kaufmann, 1988.C OMPUTATIONALD ECISION A NALYSIS206[BHM77]Bradley, S.P., Hax, A.C., and Magnanti, T.L.: Applied Mathematical Programming , Addison-Wesley, 1977.[B84]Broder, J.F.: Risk Analysis and the Security Survey , Butterworth Publishers, 1984.[B86]Brooks, R.A.: “A Robust, Layered Control System for a Mobile Robot”in IEEE Journal of Robotics and Automation , vol.2, pp.14–23, 1986.[BMM88]Budnick, F.S., McLeavey, D.W., and Mojena, R.: Principles of Operations Research in Management , 2.ed., Irwin, 1988.[CH92]Chen, S-J., and Hwang, C-L.: Fuzzy Multiple Attribute Decision Making , Lecture Notes in Economics and Mathematical Systems,vol.375, Springer-Verlag, 1992.[C83]Chvátal, V.: Linear Programming , W.H. Freeman, 1983.[CW88]Copeland, T.E., and Weston, J.F.: Financial Theory and Corporate Policy , 3.ed., Addison-Wesley, 1988.[CPS92]Cottle, R.W., Pang, J-S., and Stone, R.E.: The Linear Complementarity Problem , Academic Press, 1992.[C77]Courtney, R.H.: “Security Risk Assessment in Electronic Data Processing” in AFIPS NCC , vol.46, 1977.[D93]Danielson, M.: Implementation av ett system för beslutsanalys,WP -196, Dept. of Computer and Systems Sciences, Royal Institute of Technology, 1993.[D95]Danielson, M.: Computing Best Choices using Imprecise Information,Licentiate Thesis, Dept. of Computer and Systems Sciences, Royal Institute of Technology, 1995.[D96]Danielson, M.: DDT – the D ELTA Decision Tool, Research Report 96-026, Dept. of Computer and Systems Sciences, Royal Institute of Technology, 1996. Presented at the IIASA workshop on Advances in Methodology and Software for Decision Support Systems and Software, Laxenburg, Austria, September 1996.[D97]Danielson, M.: DDT dwish Specification, version 2.4, Dept. of Computer and Systems Sciences, Royal Institute of Technology, 1997.[DE97a]Danielson, M., and Ekenberg, L.: “Evaluating Decision Trees under Different Criteria”, 13th Int. Conf. on Multiple Criteria Decision Aids, Cape Town, 1997.[DE97b]Danielson, M., and Ekenberg, L.: Riskbedömning i vaga domäner,N UTEK Project Report, 1997.[DE97c]Danielson, M., and Ekenberg, L.: “A Framework for Analysing Decisions under Risk” to appear in European Journal of Operational Research , 1997.[DEE96]Danielson, M., Ekenberg, L., and Elgemyr, A.: “Riskanalys med DEEP -metoden” in Scandinavian Insurance Quarterly , vol.77, no.4,pp.311–324, 1996.R EFERENCES207[D67]Dempster, A.P.: “Upper and Lower Probabilities Induced by a Multivalued Mapping” in Annals of Mathematical Statistics ,vol.XXXVIII , pp.325–339, 1967.[D90]Dixon, G.: Riskanalys , SBF – Svenska Brandförsvarsföreningen, 1990.[D92]Doyle, J.: “Rationality and its Roles in Reasoning” in Computational Intelligence , vol.8, no.2, pp.376–409, 1992.[E77]Edwards, W.: “How to Use Multi-Attribute Utility Measurement for Social Decisionmaking” in IEEE Transactions on Systems, Man, and Cybernetics , vol.7, no.5, pp.326–340, 1977.[E94]Ekenberg, L.: Decision Support in Numerically Imprecise Domains,Ph.D. Thesis, Dept. of Computer and Systems Sciences, Stockholm University, 1994.[E96]Ekenberg, L.: “Modelling Decentralised Decision Making” in Proceedings of the 2nd International Conference on Multi-Agent SystemsICMAS ’96, AAAI/MIT Press, 1996.[EBD95]Ekenberg, L., Boman, M., and Danielson, M.: “A Tool for Co-ordinating Autonomous Agents with Conflicting Goals” in Proceedings of the 1st International Conference on Multi-Agent Systems ICMAS ’95,pp.89–93, AAAI/MIT Press, 1995.[ED94]Ekenberg, L., and Danielson, M.: “A Support System for Real-Life Decisions in Numerically Imprecise Domains” in Operations Research Proceedings 1994, Berlin, Germany, eds. Derigs, Bachem, and Drexl,pp.500–505, Springer-Verlag, 1994.[ED95]Ekenberg, L., and Danielson, M.: “Handling Imprecise Information in Risk Management” in Information Security – the Next Decade , eds.Eloff and von Solms, pp.357–368, Chapman & Hall, 1995.[EDB96a]Ekenberg, L., Danielson, M., and Boman, M.: “A Tool for Handling Uncertain Information in Multi-Agent Systems” in Distributed Software Agents and Applications – Proceedings of the MAAMAW ’94,eds. Perram and Müller, pp.54–62, Springer-Verlag, 1996.[EDB96b]Ekenberg, L., Danielson, M., and Boman, M.: “From Local Assessments to Global Rationality” in International Journal of Cooperative Information Systems , vol.5, nos.2&3, pp.315–331, 1996.[EDB97]Ekenberg, L., Danielson, M., and Boman, M.: “Imposing Security Constraints on Agent-Based Decision Support” to appear in Decision Support Systems International Journal , 1997.[EM92]Elgemyr, A., and Mattsson, L.: Stora säkerhetsboken , Publica, 1992.[ESF91]European Security Forum: A Risk Analysis Method which is Easy to Understand and Simple to Apply , Draft Method, 1991.[F70]Fishburn, P.C.: Utility Theory for Decision Making , John Wiley and Sons, 1970.[F81]Fishburn, P.C.: “Subjective Expected Utility: A Review of Norma-tive Theories” in Theory and Decision , vol.13, pp.139–199, 1981.C OMPUTATIONALD ECISION A NALYSIS208[F83]Fishburn, P.C.: “Transitive Measurable Utility” in Journal of Economic Theory , vol.31, pp.293–317, 1983.[F80]Freeling, A.N.S.: “Fuzzy Sets and Decision Analysis” in IEEE Transactions on Systems, Man, and Cybernetics , vol.10, no.7,pp.341–354, 1980.[F88]French, S.: Decision Theory: An Introduction to the Mathematics of Rationality , Ellis Horwood, 1988.[FH84]Fuller, R.J., and Hsia, C-C.: “A Simplified Common Stock Valuation Model” in Financial Analysts Journal, Sep.-Oct. 84,pp.49–56, 1984.[G86]Galliers, J.R.: “A Theoretical Framework for Computer Models of Cooperative Dialogue, Acknowledging Multi-Agent Conflict” in IEEE Journal of Robotics and Automation , vol.2, pp.14–23, 1986.[GN87]Genesereth, M.R., and Nilsson, N.J.: Logical Foundations of Artificial Intelligence , Morgan Kaufmann, 1987.[GD93]Gmytrasiewicz, P.J., and Durfee, E.H.: “Elements of a Utilitarian Theory of Knowledge and Action” in Proceedings of 13th IJCAI ,pp.396–402, 1993.[GT89]Goldfarb, D., and Todd, M.J.: “Linear Programming” in Optimi-zation , Handbooks in Operations Research and Management Science,vol.1, eds. Nemhauser, Rinnooy Kan, and Todd, Elsevier, 1989.[G62]Good, I.J.: “Subjective Probability as the Measure of a Non-measurable Set” in Logic, Methodology, and the Philosophy of Science ,eds. Suppes, Nagel, and Tarski, pp.319–329, Stanford University Press, 1962.[G92]Gonzaga, C.C.: “Path-Following Methods for Linear Programming”in SIAM Review , vol.14, no.2, pp.167–224, Society for Industrial and Applied Mat hemat ics, 1992.[G92b]Green, B.: “Vad kan bankerna lära sig av en entrepenör som utvecklas till organisationsforskare” in Riskbedömning – kunskap om risker ,N UTEK , Stockholm, pp.121–126, 1992.[GS82]Gärdenfors, P., and Sahlin, N-E.: “Unreliable Probabilities, Risk Taking, and Decision Making” in Synthese , vol.53, pp.361–386, 1982.[H60]Halmos, P.R.: Naive Set Theory , D. van Nostrand Co., 1960.[H88]Hamilton, G.: This is Risk Management , Chartwell-Bratt, 1988.[HM53]Hernstein, I.N., and Milnor, J.: “An Axiomatic Approach to Measurable Utility” in Econometrica , vol.21, pp.291–297, 1953.[HM84]Howard, R.A., and Matheson, J.E.: “Influence Diagrams” in Principles and Applications of Decision Analysis , eds. Howard and Matheson,vol.II, Strategic Decisions Group, Menlo Park, CA , USA , 1984.[H89]Hull, J.: Options, Futures, and Other Derivative Securities , Prentice-Hall, 1989.R EFERENCES209[H51]Hurwicz, L.: Optimality Criteria for Decision Making under Ignorance ,Cowles Commission Discussion Paper no.370, 1951.[J88]Jarrow, R.A.: Finance Theory , Prentice-Hall, 1988.[J83]Jeffrey, R.: The Logic of decision , University of Chicago Press, 1983.[K92]Keeney, R.L.: Value-Focused Thinking: A Path to Creative Decision Making , Harvard University Press, 1992.[KR76]Keeney, R.L., and Raiffa, H.: Decisions with Multiple Objectives:Preferences and Value Trade-offs , John Wiley and Sons, 1976.[K96]Kemikontoret: Riskhantering 1, Administrativ SHM -revision, 4.ed,1996.[K90]Kreps, D.M.: A Course in Microeconomic Theory , Harvester Wheatsheaf,1990.[K87]Krovak, J.: “Ranking Alternatives – Comparison of Different Methods Based on Binary Comparison Matrices” in European Journal of Operational Research , vol.32, pp.86–95, 1987.[LH94]Lai, Y-J., and Hwang, C-L.: Fuzzy Multiple Objective Decision Making ,Lecture Notes in Economics and Mathematical Systems, vol.404,Springer-Verlag, 1994.[L25]Laplace, P.: Essai Philosophique sur les Probabilites , 5.ed., Paris, 1825.(Translation published by Dover 1952.)[LF90]Lee, C.F., and Finnerty, J.E.: Corporate Finance: Theory, Method, and Applications , Harcourt Brace Jovanovich, 1990.[L59]Lehmann, E.L.: Testing Statistical Hypothesis , John Wiley and Sons,1959.[L74]Levi, I.: “On Indeterminate Probabilities” in The Journal of Philosophy , vol.71, pp.391–418, 1974.[L92]Levi, I.: “Feasibility” in Knowledge, Belief, and Strategic Interaction ,eds. Bicchieri and Chiara, Cambridge University Press, pp.1–20,1992.[LS82]Loomes, G., and Sugden, R.: “Regret Theory: An Alternative Theory of Rational Choice under Uncertainty” in The Economic Journal ,vol.92, pp.805–924, 1982.[L93]Lootsma, F.A.: “Scale Sensitivity in the Multiplicative AHP andSMART ” in Journal of Multi-Criteria Decision Analysis , vol.2,pp.87–110, 1993.[L89]Luenberger, D.G.: Linear and Nonlinear Programming , 2.ed., Addison-Wesley, 1989.[M81]Malmnäs, P-E.: From Qualitative to Quantitative Probability , Ph.D.Thesis, Almqvist & Wiksell International, 1981.[M90]Malmnäs, P-E.: Real-Life Decisions, Expected Utility, and Effective Computability, Research Report HSFR no.677/87, 1990.[M93]Malmnäs, P-E.: Festina Lente – En beslutsanalys in kärnkraftsfrågan,Research Report, Dept. of Philosophy, Stockholm University, 1993.C OMPUTATIONALD ECISION A NALYSIS210[M94a]Malmnäs, P-E.: “Towards a Mechanization of Real Life Decisions”in Logic and Philosophy of Science in Uppsala , eds. Prawitz and Westerståhl, Kluwer Academic Publishers, 1994.[M94b]Malmnäs, P-E.: “Axiomatic Justifications of the Utility Principle”in Synthese , vol.99, no.2, pp.233–249, 1994.[M96]Malmnäs, P-E.: Evaluations, Preferences, and Choice Rules, Dept. of Philosophy, Stockholm University, 1996.[MM73]Mason, R.O., and Mitroff, I.I.: “A Program for Research on Management Information Systems” in Management Science , vol.19,no.5, 1973.[M92]McClennen, E.F.: “Rational Choice in the Context of Ideal Games”in Knowledge, Belief, and Strategic Interaction , eds. Bicchieri and Chiara, pp.47–60, Cambridge University Press, 1992.[M34]Menger, K.: “Das Unsicherheitsmoment in der Wertlehre” in Zeitschrift für Nationalökonomie , vol.5, pp.49–59, 1934.[M54]Milnor, J.: “Games against Nature” in Decision Processes , eds. Thrall,Coombs, and Davis, pp.49–59, John Wiley and Sons, 1954.[N90]Neapolitan, R.E.: Probabilistic Reasoning in Expert Systems: Theory and Algorithms , John Wiley and Sons, 1990.[NM47]von Neumann, J., and Morgenstern, O.: Theory of Games and Economic Behavior , 2.ed., Princeton University Press, 1947.[N81]Newell, A.: “The Knowledge Level” in AI Magazine, Summer Issue,pp.1–20, 1981.[N86]Nilsson, N.J.: “Probabilistic Logic” in Artificial Intelligence , vol.28,pp.71–87, 1986.[OM90]Oddie, G., and Milne, P.: “Act and Value” in Theoria , vol.LVII ,pp.42–76, 1990.[O96]Olson, D.L.: Decision Aids for Selection Problems , Springer-Verlag,1996.[P91]Pearl, J.: Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference , revised 2nd printing, Morgan Kaufmann, 1991.[Q82]Quiggin, J.: “A Theory of Anticipated Utility” in Journal of Economic Behavior and Organisation , vol.3, pp.323–343, 1982.[R92]Rabinowicz, W.: “Tortuous Labyrinth: Noncooperative Normal-Form Games between Hyperrational Players” in Knowledge, Belief, and Strategic Interaction , eds. Bicchieri and Chiara, Cambridge University Press, pp.107–126, 1992.[R68]Raiffa, H.: Decision Analysis: Introductory Lectures on Choices under Uncertainty , Random House, 1968.[R78]Ramsey, F.P.: “Truth and Probability” in Foundations: Essays in Philosophy, Logics, Mathematics and Economics , ed. Mellor,pp.58–100, Routledge & Kegan Paul, 1978.R EFERENCES 211[R93]Rosenschein, J.S.: “Consenting Agents: Negotiation Mechanisms for Multi-Agent Systems” in Proceedings of 13th IJCAI , pp.792–799, 1993.[RS95]Russell, S.J., and Subramanian, D.: “Provably Bounded-Optimal Agents” in Journal of Artificial Intelligence Research , vol.2,pp.575–609, 1995.[S80]Saaty, T.L.: The Analytical Hierarchy Process , McGraw-Hill, 1980.[SAF86]SAF: Riskanalys , Näringslivets Beredskapsbyrå, 1986.[SW84]Sage, A.P., and White, C.C.: “ARIADNE : A Knowledge-Based Interactive System for Planning and Decision Support” in IEEE Transactions on Systems, Man, and Cybernetics , vol.14, no.1, 1984.[SH95]Salo, A.A., and Hämäläinen, R.P.: “Preference Programming through Approximate Ratio Comparisons” in European Journal of Operational Research , vol.82, pp.458–475, 1995.[S51]Savage, L.: “The Theory of Statistical Decision” in Journal of the American Statistical Association , vol.46, pp.55–67, 1951.[S72]Savage, L.: The Foundations of Statistics , 2.ed, Dover, 1972.[S82]Schoemaker, P.: “The Expected Utility Model: Its Variants,Purposes, Evidence, and Limitations” in Journal of Economic Literature , vol.XX , pp.529–563, 1982.[S86]Shachter, R.D.: “Evaluating Inference Diagrams” in Operations Research , vol.34, no.6, 1986.[S76]Shafer, G.: A Mathematical Theory of Evidence, Princeton University Press, 1976.[SP89]Shafer, G., and Pearl, J., eds.: Readings in Uncertain Reasoning , Morgan Kaufmann, 1989.[S76b]Simon, H.A.: Administrative Behaviour , 3.ed. The Free Press, New York, 1976.[S61]Smith, C.A.B.: “Consistency in Statistical Inference and Decision” in Journal of the Royal Statistic Society , Series B, vol.XXIII , pp.1–25, 1961.[S89–91]Statskontoret: Vägledning i ADB -säkerhet 1–8, 1989–91.[V92]Vincke, P.: Multicriteria Decision Aid , John Wiley and Sons, 1992.[W50]Wald, A.: Statistical Decision Functions , John Wiley and Sons, 1950.[W97]Walter, J.: A Decision Tool for Uncertain Decision Making, Master’s Thesis, Dept. of Numerical Analysis and Computing Science,Stockholm University, 1997.[WF82]Watson, S.R., and Freeling, A.N.S.: “Assessing Attribute Weights”in OMEGA , vol.10, pp.582–583, 1982.[WP90]Weichselberger, K., and Pöhlman, S.: A Methodology for Uncertainty in Knowledge-Based S ystems , Springer-Verlag, 1990.[W91b]Wermdalen, H.: Securitas – Säkerhetsboken 1992, Studentlitteratur,1991.C OMPUTATIONALD ECISION A NALYSIS212[W84]Wrede, R.: “The SBA Method: A Method for Testing Vulnerability”in Proc. IFIP/SEC '84, pp.313–320, 1984.[W91]Wright, M.H.: “Interior Methods for Constrained Optimization”in Acta Numerica , pp.341–407, 1991.[Y87]Yaari, M.: “The Dual Theory of Choice under Risk” in Econometrica ,vol.55, pp.95–115, 1987.[Z96]Zilberst ein, S.: “Using Anyt ime Algorit hms in Int elligent Syst ems”in AI Magazine , Fall 1996, pp.73–83, 1996.[ZZG84]Zimmermann, H.J., Zadeh, L.A., and Gaines, B.R.: Fuzzy Sets and Decision Analysis, TIMS Studies in the Management Sciences, vol.20,North-Holland, 1984.。
Generalizing Data in Natural LanguageRyszard S.Michalski12and Janusz Wojtusiak11Machine Learning and Inference Laboratory,George Mason University,Fairfax,VA22030,USA2Institute of Computer Science,Polish Academy of Science,Warsaw,PolandAbstract.This paper concerns the development of a new direction inmachine learning,called natural induction,which requires from computer-generated knowledge not only to have high predictive accuracy,but alsoto be in human-oriented forms,such as natural language descriptionsand/or graphical representations.Such forms facilitate understandingand acceptance of the learned knowledge,and making mental modelsthat are useful for decision making.An initial version of the AQ21-NIprogram for natural induction and its several novel features are brieflydescribed.The performance of the program is illustrated by an exampleof deriving medical diagnostic rules from micro-array data.1IntroductionMost of machine learning research has been striving for achieving high predictive accuracy of knowledge learned from data,but has not paid much attention to the understandability and interpretability of that knowledge.This is evidenced by the fact that research papers on different learning methods,including learning decision trees,random forests,decision rules,ensemblies,neural nets,support vector machines,etc.,typically list only predictive accuracies obtained by the reported and compared methods(e.g.,[1]),but very rarely present actual knowl-edge learned.While predictive accuracy of inductively acquired knowledge is obviously im-portant,for many applications it is imperative that computer-generated knowl-edge is in the forms facilitating its understanding and making mental models of it by an expert.Suchfields include,for example,medicine,bioinformatics, agriculture,social sciences,economy,business,archeology,defense,and others. Although the need for understandability of computer-generated knowledge has been indicated for a long time(e.g.,[7],[13]),research on this topic has been inadequate.The main reason for this situation may be that understandability and interpretability of knowledge is subjective and difficult to measure.This paper concerns the development of a new direction in machine learn-ing,called natural induction,which strives to achieve high understandability and interpretability of computer-generated knowledge by learning it and pre-senting it in the forms resembling those in which people represent knowledge. Such forms include natural language descriptions and simple graphical repre-sentations.To serve this objective,we employed attributional calculus[8]as alogic and knowledge representation for learning.Attributional calculus combines selected features of propositional,predicate and multi-valued logics,and intro-duces several new constructs formalizing relevant features of natural language. We developed algorithms for learning attributional rules with these constructs, and also for transforming these rules into simple natural language descriptions. These algorithms have been implemented in the AQ21-NI program,briefly,NI, whose selected features are described in this paper.2Brief Overview of Natural InductionThe natural induction methodology for learning natural language descriptions from data involves three stages of processing.Thefirst stage induces formal rules in attributional calculus.Such rules are more expressive than standard decision rules in which conditions are limited to¡attribute relation value¿forms and are also closer to equivalent natural language descriptions.The second stage transforms learned attributional rules into logically equivalent and grammati-cally correct natural language descriptions.The third stage employs cognitive constraints and relevant background knowledge to improve the descriptions’in-terpretability and to derive additional implications from them that are useful for decision making.This paper concerns thefirst two stages.The third stage is under development.Let us start by briefly characterizing the general task addressed by thefirst stage.The goal of this stage is to take a set of data points(training examples) that exemplify decision classes C1,...,C k,and relevant background knowledge, and induce hypotheses,H1,...,H k that generally describe these classes and op-timize a multi-criterion measure of of description quality.In the method imple-mented in the AQ21-NI program,the generated hypotheses are different forms of attributional rules.Adopting formalism presented in[8],the basic form of an attributional rule is:CONSEQUENT<=P REMISE(1)where CONSEQUENT and PREMISE are conjunctions of attributional condi-tions,that are formal equivalents of simple natural language statements.Here is an example of a basic attributional rule:[Task_to_do=run_experiments]<=[Day=weekday]&[#tissue-samples-to-analyze=2..7]& [Tests-to-perform:PAP&estradiol_level]&[available-lab=lab1v lab3]The second stage transforms the learned attributional rules into equivalent and grammatically correct natural language descriptions.For example,the above rule is translated to the following natural language description:”The task is to run experiments,if the day is weekday,the number of tissue samples to analyze is between2and7,the tests to perform are PAP and estradiol level,and the available lab is lab1or lab3.”As one can notice,the above natural language description closely corresponds to the attributional rule from which it was derived.The”weekday”is a program-abstracted value of the structured attribute”day”(the domain of a structured at-tribute is a hierarchy).The attribute”tests-to-perform”is a compound attribute whose legal values are internal conjunctions of values of constituent attributes (an internal conjunction binds atoms rather than statements).As this example shows,attributional rules significantly extend standard de-cision rules whose conditions are limited to the form[ATTR REL VAL],where ATTR is a single attribute,REL is=,≤or≥and VAL is an attribute value. It addition to constructs presented in this example,attributional rules may in-volve also conditions with count attribute that counts the number of statements that are true in a given set,or counts the number properties satisfying a given condition.Expressions with count attributes can be viewed as a special case of statements in the second order predicate calculus(Section2.4).Attribut-ional rules may also include exception clause(Section2.2),and several other forms that resemble those used by people in natural language descriptions(e.g., provided-that clause).2.1The Q(w)Criterion of Description OptimalityThefirst stage integrates the well-known separate and conquer algorithm AQ for learning consistent and complete rules(e.g.,[6]or[7]),with an algorithm for discovering patterns from data,and with procedures for learning new con-structs briefly mentioned above.These constructs are described in more detail in Sections2.2to2.5.The implementation of this stage is based on the AQ21learn-ing and pattern discovery system[20]that enhances the basic AQ-type learning method by a number of new features.Among the new features is the ability of the program to work in either Theory Formation(TF)or Pattern Discovery(PD)mode,which is controlled by the”mode”parameter.The TF mode can generate several different types of optimized complete and consistent descriptions of training examples,such as attributional rules without or with exception clauses(Section2.2).The rules optimize a user-defined multi-criterion measure of rule optimality,LEF(e.g.,[6] or[7]).The PD mode searches for patterns or approximate descriptions that maxi-mize a description optimality criterion defined as:Q(R,w)=cov w∗config1−w(2) where cov=p/P and config=((p/(p+n))-(P/(P+N)))*(P+N)/N are measures of coverage and confidence gain,respectively,of the rule R,and w is a user-controlled parameter.Here,p and n are the numbers of positive and negative examples covered by R,and P and N are the numbers of positive and negative examples in training dataset,respectively.The confidence gain captures the increase of confidence in the rule in relation to confidence in decisions made according to their prior probabilities.As one can see,the criterion Q allows trading inconsistency(n=0)for an increase in rule coverage.2.2Learning Descriptions with Exception ClausesExceptions are commonly used by humans when describing rarely occurring anomalies that are inconsistent with a given rule or a theory.It is not unusual that an approximate description of observations can be very simple,but a perfect description,fully consistent with all observations,would be significantly more complex.In such cases,it may be useful to learn rules with exception clauses,also called censored rules(e.g.,[10],[8],[17]).AQ21-NI can be set to learn censored rules in the form:CONSEQUENT<=P REMISE|EXCEP T ION(3) where|is an exception operator,and EXCEPTION is either a conjunctive attributional description(an exception clause)or a list of examples constituting exceptions to the basic rule.The rule is read:If PREMISE is true then assert CONSEQUENT,except when EXCEPTION is true.In PD mode,where inconsistency is allowed,the program learns basic rules that maximize Q(R,w),and then adds to them exception clauses.The latter are generated by re-applying the AQ algorithm to the examples covered by the rule to describe negative examples covered by this rule.In TF mode,where consistency has to be guaranteed,learning censored rulesfirst involves creating basic rules and an exception list for each of them.Such a list contains examples that would introduce a significant complexity,if the rule was transferred into an expression consisting of fully consistent rules,and only if the number of examples on the exception list is significantly smaller than the number of positive examples the rule covers.If all of the exceptions on a list can be characterized by one conjunctive statement,then it is used as the EXCEPTION clause;otherwise, EXCEPTION is an explicit list of exceptions.To illustrate differences between basic and censored rules,let us consider a simple problem of learning descriptions for”friendly”robots from their examples and counter examples.When asked to produce basic rules,AQ21-NI created two rules(the premise of each rule is preceded by<=).[Robot=friendly]<=[Holding=book:4,4]&[Size=small..medium:6,6]:p=4,n=0<=[Holding=book v flag:6,8]&[~Antennas:3,9]&[Size=small..medium:6,6]:p=3,n=0The numbers inside conditions,after a”:”,represent their positive and nega-tive coverage of the condition,respectively;parameters p and n after each rule represent the number of positive and negative examples covered by the rule, respectively.When asked to produce censored rules,the program generated a single rule that also covers completely and consistently the training examples:[Robot=friendly]<=[Holding=book v flag:6,8]&[Size=small..medium:6,6]|_[Holding=flag:2,3]&[Antennas:1,5]:p=6,n=0When transformed into an equivalent natural language expression,the above censored rule becomes:”A robot is friendly,if it is holding a book or a balloon,and its size is be-tween small and medium,inclusively,except when it is holding aflag and has antennas.”Note that the numbers of positive examples covered by conditions in the EX-CEPTION clause are significantly smaller than those covered by the PREMISE, and the numbers of negative examples covered exceed the numbers of negative examples covered by rule conditions.2.3Learning Descriptions with Compound AttributesOne of the novel features of AQ21-NI is that it implements compound attributes that facilitate learning natural language descriptions of objects,or their compo-nents that require different attributes to describe them.Consider,for example, a description of weather in the style of standard propositional logic:[Windy=yes]&[Cloudy=yes]&[Humid=not]Using a compound attribute,such a description would be expressed as: [Weather:windy&cloudy¬ humid]that resembles the equivalent natural language statement:”Weather is windy, cloudy and not humid.”In this example,”weather”is a compound attribute, and windy,cloudy,and humid are values of its constituent attributes[8].Learn-ing expressions with compound attributes is done by learning rules using con-stituent attributes,and then transforming appropriate groups of attributes into compound forms.2.4Learning Descriptions with Counting AttributesIn some applications,in particular,in medicine,it is not unusual that a medical decision(e.g.,diagnosis)is made on the basis of counting of number features, (e.g.,symptoms),observed in the patient,and comparing it with a threshold. If the number of symptoms exceeds the threshold,the disease is implicated.To illustrate such a case by a real world example,consider a problem of classifying the severity of prostate cancers in terms of three known risk factors:Factor1:PSA≥10ng/ml(”PSA”measures the amount of prostate specific antigen)Factor2:Gleason’s score≥7(”Gleason’s score”measures the cancer cells’ab-normality)Factor3:Stage≥T2b(”Stage”measures the level of disease development).Based on these factors,prostate cancer patients are classified into four cate-gories,representing an increasing severity of their disease:Category is1,if no factors are present;Category is2,if one factor is present; Category is3,if two factors are present;Category is4,if all three factors are present.This classification was obtained from Dr.P.Koutrovalis,Director of URO-Radiology Prostate Institute in Washington,ing attributional rules,the above classification schema can be represented by four attributional rules: [Category=1]<=[count(PSA≥10ng/ml,Gleason’s≥7,Stage≥T2b)=0] [Category=2]<=[count(PSA≥10ng/ml,Gleason’s≥7,Stage≥T2b)=1] [Category=3]<=[count(PSA≥10ng/ml,Gleason’s≥7,Stage≥T2b)=2] [Category=4]<=[count(PSA≥10ng/ml,Gleason’s≥7,Stage≥T2b)=3] where count(S1,S2,..,Sn)is a derived attribute that counts the number of sentences between the parentheses that are true.To express the above classifica-tion schema by a decision tree or standard decision rules would require a more complex and more difficult to interpret structure,for example,a decision tree with eight leaves and seven internal nodes,or eight standard rules.Expressions with a count attribute are generalizations of the so-called n-of-m relations(stating that n of m binary attributes are true in a description)[15]. AQ21-NI can learn,however,not only n-of-m special cases,but more general expressions that involve both count attributes and other conditions,for example: [DiseaseState=severe]<=[count(C1,C5,C8)≥2]&[Abnormality-type=A v C] As one can see,the attributional rules can express quite elaborate conditions and are closely related to the equivalent natural language descriptions.The latter feature makes them easy to translate into such descriptions.2.5Learning Optimized Sets of Alternative ClassifiersFrom any non-trivial set of concept examples,it is usually possible to generate many alternative inductive generalizations of these examples.Such alternative hypotheses can be useful in a variety of practical applications.For example, in medicine,it may be desirable to generate alternative explanations of the symptoms to protect a doctor from overlooking a rare disease.AQ21-NI seeks a collection of alternative classifiers that optimizes a user-defined multi-criterion measure.Here,a classifier is a collection of attributional rulesets,where each ruleset is associated with one value,e.g.,one disease,in the domain of the output attribute,e.g.,diagnosis(for more explanation,consult[8]).The purpose of optimizing the collection is to include in it,for example,alternative classifiers that are maximally different from each other.For example,for the robots problem presented above,one execution of AQ21-NI generates two alternative rulesets for the class”friendly”at thefirst stage of processing:Classifier1:[robot=friendly]<=[holding=book]&[size=small..medium]:p=4,n=0<=[holding=book v flag]&[antenas=no]&[size=small..medium]:p=3,n=0Classifier2:[class=friendly]<=[holding=book]&[size=small..medium]:p=4,n=0<=[holding=book v flag]&[size=medium]:p=4,n=0<=[holding=flag]&[size=small]&[hands=yes]:p=1,n=0.The algorithm for creating optimized collections of alternative classifiers is described in[9].3Generating Natural Language DescriptionsThe second stage involves transforming the learned attributional rulesets into grammatically correct natural language descriptions.This task is done according to the following hard-coded rules:1.Attribute names used in the rules are translated onto their natural languageequivalences provided by the user.2.Symbols”=”and”:”in rule conditions used with regular and compoundattributes are translated into the word”is.”Symbols”>”,”<”are translated into”greater than”,”smaller than,”and symbols”≥”and”≤”are translated into”at least,”and”at most”.3.Attribute values connected in a condition by internal disjunction or internalconjunction are separated by a comma,except for the last value that is separated by an”or,”or”and”,respectively.Range expressions”val1..val2”are translated into a statement”between val1and val2,inclusively”.4.Conditions with a count attribute are transformed according to a template.Ifthe count refers to statements,then the condition is transformed into”The number of true statements on the list”,followed by the list of conditions transformed to natural language,”is”,and followed by the value indicated in the count condition.If the count refers to attributes,then the condition is transformed to a statement:”The number of attributes on the list L whose values are Rel is Val”,where L,Rel and Val are indicated in the condition.For example,[count(x1,x3,x5,x8>3)=2]is transformed into”The number of attributes on the list(x1,x3,x5,x8)whose values are greater than3is2.”5.If there is more than one,but at most three implications”<=”after the rulehead,that is,the rule consequent(this number is a modifiable parameter), they are replaced by an”or.”To reflect the spirit of cognitive aspects of attributional calculus,if there are more”<=”than allowed by the parameter, they are transformed into a sentence:”The strongest rule implying<head condition>is<natural language version of the strongest rule>.Other rules also implying the<head condition>are<natural language representation of the remaining rules>”.The rules’strength is determined according to a user-defined criterion,such as Q(w)(default),coverage,confidence,etc. 6.Numbers p and n listed at the end of a rule are translated to natural lan-guage byfilling a template:”The rule is satisfied by p positive and n negative training examples.”This statement may be followed by lists of these exam-ples.Similarly,a template is used to translate the weights associated with each condition in a rule.To obtain a satisfactory natural language representation of attributional rule-sets,it is important to appropriately name attributes and their legal values in preparing the input to the program.4An Example of Applications in MedicineThis example describes an application of AQ21-NI to the problem of diagnos-ing medulloblastoma from patients’gene micro-arrays(representing degrees of expressions of patients’genes).Medulloblastoma is a highly invasive primitive neuroectodermal tumor of the cerebellum and the most common malignant brain tumor of childhood.The data for this application were obtained from Gene Ex-pression Omnibus,available from /geo.The original gene micro-array data consists of46records,split into two groups:20and26records, describing patients with metastatic and non-metastatic tumors,respectively. Each record registers values of2059real-valued attributes(representing gene expressions).In the experiments we obtained16and12unique examples of metastatic and non-metastatic tumors,respectively[11].From these examples,AQ21-NI at thefirst stage generated two simple rules for diagnosing metastatic tumor that require measuring only four gene expres-sions:[Cancer=metastatic]<=[Gene-1611-expression<=100.9:18,8]&[Gene-1036-expression=-41.76..160.8:18,20]&[Gene-914-expression<=21.5:20,15]:p=16,n=0<=[Gene-1783-expression>=96.6:6,0]:p=6,n=0An equivalent natural language description is:”Cancer is metastatic,if gene 1611expression is at most100.9,gene1036expression is between-41.76and 160.8,and gene914expression is at most21.5,or gene1783expression is at least96.6.”When the experiment was performed using5-fold cross-validation,the rules obtained by AQ21-NI had predictive accuracy about95%.It is noteworthy that in the experiments that inspired our work in this domain[5],the authors devel-oped a neural net that requires measuring80genes,and its reported predictive accuracy was about72%.Thus,their result is not only less accurate than that obtained by AQ21-NI,but is also a significantly more complex,as it requires measuring expression of many more genes.Moreover,it is a black box solution that is very difficult to interpret.5Relation to Other WorkThefirst phase of AQ21-NI is related to programs learning standard decision rules.Among such programs are CN2[2],C4.5[14],RIPPER[3],programs us-ing rough-set theory approach,e.g.,[12]and[4],programs learning fuzzy rules,e.g.,[19],and those applying evolutionary computation,e.g.[16].Because stan-dard decision rules have a relatively low expression power,these programs cannot learn more expressive attributional rules that are learned by AQ21-NI,and have fewer capabilities.To the best of author’s knowledge,AQ21-NI is the only pro-gram currently in existence that has such a large number of different capabilities integrated in one program.Also,the authors are not aware of any existing rule learning program that performs the second stage of learning,that is,generates natural language concept descriptions.Work on this stage concerns generating natural language descrip-tions from logical-style rules.The task of generating natural language descrip-tions is usually addressed from two different perspectives,the template-based, which maps non-linguistic input directly to natural language(without interme-diate representations),and the standard method,which builds sentences through a semantic analysis of the text being generated[18].As was mentioned earlier,attributional calculus facilitates learning of richer and frequently also simpler generalizations of examples than representations based standard decision rules.The cost for this advantage is,however,a signifi-cantly higher complexity of the learning algorithm,and,consequently,a longer time of its execution.Due to the great progress in increasing speed of modern computers,the second issue is of decreasing importance.In our experiments, AQ21-NI has proven to be quite efficient.An earlier version of AQ learning program was effectively applied to problems with millions of training examples. 6SummaryNatural induction aims at creating knowledge from data that is not only accu-rate but also easy to understand and interpret.The latter objective is achieved byfirst learning expressions in attributional calculus that adds to standard logic several new constructs,and then transforming the learned descriptions into equivalent natural language descriptions.A methodology for natural induc-tion has been implemented in the AQ21-NI rule learning program.The program seamlessly integrates several new features that include learning in two modes-theory formation and pattern discovery,learning with compound attributes, learning censored rules and learning optimized collections of alternative clas-sifiers.Due to space limitations,the paper includes only very brief descriptions of these features.More detailed descriptions are in publications downloadable from .An application of AQ21-NI to a problem in bioin-formatics produced a hypothesis that a medical expert evaluated as having an important medical value,because it suggests adjusting thresholds in the cur-rently used diagnostic procedure.The ability of AQ21-NI to produce natural language descriptions makes it attractive for application domains in which understandability of computer-generated knowledge is highly important,such as medicine,bioinformatics,so-ciology,psychology,economy,business,archeology,civil engineering,and others.References1.Caruana R.and Niculescu-Mizil,A.:An Empirical Comparison of Supervised Learn-ing Algorithms.Proceedings of the23rd Intl Conference on Machine Learning(2006) 2.Clark,P.and Niblett,T.:The CN2Induction Algorithm.Machine Learning.3(1989)261-2893.Cohen,W.:Fast Effective Rule Induction.Proc.of the12th Intl Conference onMachine Learning(1995)4.Grzymala-Busse J.W.:Rough Set Strategies to Data with Missing Attribute Values.Proc.of the Workshop on Found.and New Directions in Data Mining(2003)5.MacDonald,T.J.,Brown,K.,LaFleur,B.,Paterson,K.,Lawlor,C.,Chen,Y.,Packer,R.,Cogen,P.and Stephan,D.:Expression Profiling of Medulloblastoma: PDGFRA and the RAS/MAPK Pathway as Therapeutic Targets for Metastatic Disease.Nature Genetics.29,(2001)143-1526.Michalski,R.S.:AQVAL/1–Computer Implementation of a Variable-Valued LogicSystem VL1and Examples of its Application to Pattern Recognition.Proceedings of the First International Joint Conference on Pattern Recognition(1973)pp.3-17 7.Michalski,R.S.:A Theory and Methodology of Inductive Learning.Artificial Intel-ligence(1983)111-1618.Michalski,R.S.:ATTRIBUTIONAL CALCULUS:A Logic and Representation Lan-guage for Natural Induction.Reports of the Machine Learning and Inference Lab-oratory MLI04-2.George Mason University(2004)9.Michalski,R.S.:Generating Alternative Hypotheses in AQ Learning.Reports of theMachine Learning and Inference Laboratory MLI04-6.George Mason Univ.(2004) 10.Michalski,R.S.and Winston,P.H.:Variable Precision Logic.Artificial IntelligenceJournal29,(1986)121-14611.Michalski,R.S.,Kaufman,K.,Pietrzykowski,J.,Wojtusiak,J.,Mitchell,S.andSeeman,W.D.:Natural Induction and Conceptual Clustering:A Review of Applica-tions.Reports of the Machine Learning and Inference Laboratory MLI06-3.George Mason University(2006)12.Pawlak,Z.:Rough Sets:Theoretical Aspects of Reasoning about Data.KluwerPublishers(1991)13.Pazzani,M.J.:Knowledge discovery from data?.IEEE Intelligent Systems.10-13,March/April(2000)14.Quinlan,J.R.:C4.5Systems for Machine Learning.Morgan Kaufmann Publ.(1993)15.Sebag,M.:Constructive Induction:A Version Space-based Approach.Proceedingsof the16th International Joint Conference on Artificial Intelligence,IJCAI99(1999) 16.Setzkorn C.and Paton R.C.:On the use of multi-objective evolutionary algorithmsfor the induction of fuzzy classification rule systems.Biosystems81,2(2005)17.Suzuki E.and Zytkow,J.M.:Unified algorithm for undirected discovery of excep-tion rules.International Journal of Intelligent Systems20,6(2005)673-69118.Van Deemter,K.,Theune,M.,and Krahmer,E.:Real vs.Template-Based NaturalLanguage Generation:A false Opposition?:Computational Linguistics31,1(2005) 19.Van Zyl,J.and Cloete,I.:Simultaneous Concept Learning of Fuzzy Rules.Pro-ceedings of the Fifteenth European Conference on Machine Learning(2004)20.Wojtusiak,J.,Michalski,R.S.,Kaufman,K.and Pietrzykowski,J.:The AQ21Natural Induction Program for Pattern Discovery:Initial Version and its Novel Features.Proceedings of The18th IEEE International Conference on Tools with Artificial Intelligence,Washington,D.C.(2006)。