CS276BWebSearchandMiningWinter2005Lecture8InformationExtractionIIFeb1,2005(includesslidesborrowedfromBBN,DavidBlei,NickKushmerick,allum,andRayMooney)1InformationExtractionLesliePackKaelbling,:,pages237-285,May1996. Author Title JournalReferencesHeaders2PlanforIEFirstclassIntroductiontotheIEproblemWrappersandWrapperInductionTraditionalNLP-basedIE:petitionsTodayPatternLearningSystems:RapierProbabilisticsequencemodels:uratepatternsforeachslotforeachdomain():Buildatrainingsetofdocumentspairedwithhuman--LearningSystemsPros::Annotatedtrainingdata,andlotsofit,’tnecessarilybetterorcheaperthanhand-builtsol’nExamples:Riloffetal.,AutoSlog,SoderlandWHISK(UMass);(UTexas);Ciravegna(Sheffield)Learnlexico-syntacticpatternsfromtemplatesTrainerDecoderModelLanguageInputAnswersAnswersLanguageInput5Rapier[Califf&Mooney,AAAI-99]Rapierlearnsthreeregex-stylepatternsforeachslot:Pre-fillerpatternFillerpatternPost-onstraints.(Seealso:SIFT[Milleretal,MUC-7];SRV[Freitag,AAAI-98];Whisk[Soderland,MLJ-99].)RAPIERrulesforextracting“transactionprice”“…paid$pany…”“…soldtothebankforanundisclosedamount…”“…paidHoneywellanundisclosedprice…”6Part-of-speechtags&lassesPartofspeech:syntacticroleofaspecificwordnoun(nn),propernoun(nnp),adjectve(jj),adverb(rb),determiner(dt),verb(vb),“.”(“.”),…NLP:Well-knownalgorithmsforautomaticallyassigningPOStagstoEnglish,French,Japanese,…(>95%accuracy)lasses:Synonymsorotherrelatedwords“Price”class:price,cost,amount,…“Month”class:January,February,Mar
CS276B Web Search and MiningWinter 2005 来自淘豆网www.taodocs.com转载请标明出处.