计算机专业英语-专业知识

NewTechnologiesAbbreviations•3D—3Dimension•ACM—AssociationforComputingMachinery•AI—ArtificialIntelligence•ALT—AdvancedLearningTechnology•API—ApplicationProgrammingInterface•CAD—ComputerAidedDesign•CBIR—Content-BasedImageRetrieval•CBVIR—Content-BasedVisualInformationRetrieval•QBIC—QuerybyImageContent•CG—ComputerGraphicsAbbreviations•CGI—CommonGatewayInterface•DVD—DigitalVideoDisk•GIS—GeographicInformationSystem•GPS—GlobalPositioningSystem•HCI—HumanComputerInteraction•HDTV—High-DefinitionTelevision•HTML—HypertextMarkupLanguage•IASTED—InternationalAssociationofScienceandTechnologyforDevelopment•IEEE—InstituteofElectricalandElectronicsEngineersAbbreviations•ISO—InternationalOrganizationforStandardization•LBS—Location-basedServices•MD—MolecularDynamics•MIS—ManagementInformationSystem•NLP—NaturalLanguageProcessing•NMR—NuclearMagneticResonance•PDF—PortableDocumentFormat•Ph.D.—DoctorofPhilosophy•RDF—ResourceDescriptionFramework•SDK—SoftwareDevelopmentKitAbbreviations•TIFF—TaggedImageFileFormat•URL—UniformResourceLocator•VLE—VirtualLearningEnvironment•VRML—VirtualRealityModelingLanguage•W3C—WorldWideWebConsortium•XML—eXtensibleMarkupLanguage•ERP—Enterpriseresourceplanningsoftware•e.g.—exempligratia•i.e.—idest6NaturalLanguageProcessing(NLP)7ThinkAbout…Whatresearchareadoesnaturallanguageprocessingbelongto?Whatproblemsdoesittrytoresolve?Whatdonatural-language-generationsystemsandnatural-language-understandingsystemsdorespectively?Whatproblemsseemdifficultinnaturallanguageprocessing?WhatarethemajortasksinNLP?8WhatisNaturalLanguageProcessing•Naturallanguageprocessing(NLP)isasubfieldofartificialintelligenceandcomputationallinguistics.Itstudiestheproblemsofautomatedgenerationandunderstandingofnaturalhumanlanguages.9WhatisNaturalLanguageProcessing•Natural-language-generationsystemsconvertinformationfromcomputerdatabasesintonormal-soundinghumanlanguage.Natural-language-understandingsystemsconvertsamplesofhumanlanguageintomoreformalrepresentationsthatareeasierforcomputerprogramstomanipulate.10Difficulties•SpeechSegmentation•TextSegmentation•WordSenseDisambiguation–Manywordshavemorethanonemeaning•SyntacticAmbiguity–Thegrammarfornaturallanguagesisambiguous•ImperfectorIrregularInput–Foreignorregionalaccentsandvocalimpedimentsinspeech;typingorgrammaticalerrors•SpeechActsandPlansSpeechSegmentation•Inmostspokenlanguages,thesoundsrepresentingsuccessivelettersblendintoeachother,sotheconversionoftheanalogsignaltodiscretecharacterscanbeaverydifficultprocess.TextSegmentation•SomewrittenlanguageslikeChinese,JapaneseandThaidonothavesingle-wordboundarieseither,soanysignificanttextparsingusuallyrequirestheidentificationofwordboundaries,whichisoftenanon-trivialtask.13MajorTasksinNLP•Automaticsummarization•ForeignLanguageReadingAid•ForeignLanguageWritingAid•Informationextraction•Informationretrieval•Machinetranslation•Namedentityrecognition14MajorTasksinNLP•Naturallanguagegeneration•OpticalCharacterRecognition•Questionanswering•Speechrecognition•Spokendialoguesystem•Textsimplification•Texttospeech•Text-proofingContent-basedImageRetrieval（CBIR)ThinkAbout…Whatiscontent-basedimageretrieval?Whatdoestheterm“content”mean?Whatarelowlevelimageretrieval,regionbasedimageretrieval,andsemanticimageretrievalrespectively?Fromthehistoricoverview,howhasCBIRevolved?Whatdoesitmeanbymultimediainformationretrieval?WhatResearchAreasAreInvolvedIn?•Computervision,patternrecognition,imageprocessing,datamining,machinelearning,human-computerinteraction,artificialintelligence•Application:digitalmuseum/libraries,safetyofsociety,image/videocopydetection,GIS,medicine,education,entertainment,•Searchfordigitalimagesinlargedatabases–Firstgeneration:laborious,subjectiveMetadata(captionsorkeywords)→Image–Secondgeneration(content-based):objectiveImagecontents→Image–Currentway(semantic):subjective+objectiveImagecontents+semanticfeature→Image–Ourway:Imagecontents+semanticfeature+keywords/captions→ImageSemanticgapWhatIsContent-basedImageRetrieval•“Content-based”meansthatthesearchwillanalyzetheactualcontentsoftheimage.Theterm“content”inthiscontextmightrefercolors,shapes,textures,oranyotherinformationthatcanbederivedfromtheimageitselfLowLevelImageRetrieval•Color–Examiningimagesbasedonthecolorstheycontainisoneofthemostwidelyusedtechniquesbecauseitdoesnotdependonimagesizeororientation.Colorsearcheswillusuallyinvolvecomparingcolorhistograms,thoughthisisnottheonlytechniqueinpracticeColorSpaceRGBLightness(亮度，即明暗）Hue（色调，即光的颜色）Saturation（饱和度，即颜色的深浅）Chrominance(色度）65.738129.05725.06416137.94574.494112.439128256112.43994.15418.285128YRCbGCrBLowLevelImageRetrieval•Shape–Shapedoesnotrefertotheshapeofanimagebuttotheshapeofaparticularregionthatisbeingsoughtout.Shapeswilloftenbedeterminedfirstapplyingsegmentationoredgedetectiontoanimage.Insomecasesaccurateshapedetectionwillrequirehumaninterventionbecausemethodslikesegmentationareverydifficulttocompletelyautomate.EdgeDetectionLowLevelImageRetrieval•Texture–Tex