资源描述:
《web as corpus--special issue.pdf》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、IntroductiontotheSpecialIssueontheWebasCorpusAdamKilgarriff∗GregoryGrefenstette†LexicographyMasterClassLtd.andITRIClairvoyanceCorporationUniversityofBrightonTheWeb,teemingasitiswithlanguagedata,ofallmannerofvarietiesandlanguages,invastquantityandfreely
2、available,isafabulouslinguistsplayground.ThisspecialissueofComputationalLinguisticsexploreswaysinwhichthisdreamisbeingexplored.1.IntroductionTheWebisimmense,free,andavailablebymouseclick.Itcontainshundredsofbillionsofwordsoftextandcanbeusedforallmanner
3、oflanguageresearch.Thesimplestlanguageuseisspellchecking.Isitspeculaterorspeculator?Googlegives67fortheformer(usefullysuggestingthelattermighthavebeenintended)and82,000forthelatter.Questionanswered.Languagescientistsandtechnologistsareincreasinglyturni
4、ngtotheWebasasourceoflanguagedata,becauseitissobig,becauseitistheonlyavailablesourceforthetypeoflanguageinwhichtheyareinterested,orsimplybecauseitisfreeandinstantlyavailable.ThemodeofworkhasincreaseddramaticallyfromastandingstartsevenyearsagowiththeWeb
5、beingusedasadatasourceinawiderangeofresearchactivities:Thepapersinthisspecialissueformasampleofthebestofit.Thisintroductiontotheissueaimstosurveytheactivitiesandexplorerecurringthemes.WefirstconsiderwhethertheWebisindeedacorpus,thenpresentahistoryofthet
6、hemeinwhichweviewtheWebasadevelopmentoftheempiricistturnthathasbroughtcorporacenterstageinthecourseofthe1990s.WebrieflysurveytherangeofWeb-basedNLPresearch,thenpresentestimatesofthesizeoftheWeb,forEnglishandforotherlanguages,andasimplemethodfortranslati
7、ngphrases.NextweopenthePandora’sboxofrepresentativeness(concludingthattheWebisnotrepresentativeofanythingotherthanitself,butthenneitherareothercorpora,andthatmoreworkneedstobedoneontexttypes).Wethenintroducethearticlesinthespecialissueandconcludewithso
8、methoughtsonhowtheWebcouldbeputatthelinguist’sdisposalrathermoreusefullythancurrentsearchenginesallow.1.1IstheWebaCorpus?ToestablishwhethertheWebisacorpusweneedtofindout,discover,ordecidewhatacorpusis.McEneryandWilson(1996,page21)sayInpr