资源描述:
《Business models for Dictionaries and NLP》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、BusinessmodelsforDictionariesandNLPAdamKilgarriNovember13,1998AbstractNLPneedsdictionaries,anddictionary-makerscanuseNLPtomakebetterdictio-naries,sothereisgreatpotentialforsynergybetweenthetwoactivities.Todate,therehasbeenonlyverylimitedcollaboration.Thetworeasonsforthisare(a)dic
2、tionarypublishers'concernsregardingintellectualproperty,and(b)thedierentlanguagesthatlexicographersandNLPresearchersspeak.InthispaperIpresentamodelforovercomingtherstandsuggestssomestrategiesforthesecond.1IntroductionNLPneedsdictionaries,anddictionary-makerscanuseNLPtomakebetter
3、dictionaries,so1thereisgreatpotentialforsynergybetweenthetwoactivities.ThereisamplemotivationforNLPtocourtdictionarypublishers,andviceversa.Todate,NLPresearchhasuseddictionariesanddictionarieshaveusedNLP,butthetwoprocesseshavenotbeenbroughttogether.TheNLPthathasgoneintomakingdicti
4、onarieshasnotbeentheNLPthatwasbasedonanearlierversionofthesamedictionary(orindeedof2anypublisheddictionaries).TheNLPhaseitherbeendictionary-independentifbroughtinfromoutside,ordevelopedinhouse.WhileNLPgroupshavemadeinnumerablecorrections,improvements,additionsandextensionstothedic
5、tionarydatabasestheyhavelicensedfrompublishers,thesechangeshaveneverbeenusedbythepublishertoimprovethenextprintingorthenexteditionofthedictionary.Therehavebeentworeasonsforthisfailureofsynergy.Firstly,dictionarypublishersareconcernedaboutanumberofquestionsrelatingtointellectualpro
6、perty.Secondly,lexicog-raphersandNLPresearchersspeakverydierentlanguages,soitisnotstraightforwardtodeveloptheatmosphereoftrustinwhichthelexicographersfullyunderstandwhattheNLPresearchhasdoneorbelievethattheNLP`enhancements'totheirdictionaryareusefultothem.1NLP(NaturalLanguageProc
7、essing)isheretakentodescribeallthosetechnologiesthatmanipulatehumanlanguageinputsbycomputer,examplesbeingautomaticpart-of-speechtagging,parsing,concordancing,machinetranslation,informationextractionandtextgeneration.`Languageengineering"and`computationallinguistics'arenear-synonym
8、s.2SophisticatedNLPsoftwarethatha