欢迎来到天天文库
浏览记录
ID:349099
大小:64.50 KB
页数:12页
时间:2017-07-26
《一个以概念演算法为基础的方法为地名消除歧义【外文翻译】》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、本科毕业设计(论文)外文翻译原文:Aconceptualdensity-basedapproachforthedisambiguationoftoponymsNowadays,ahugequantityofinformationisstoredindigitalformat.Agreatportionofthisinformationisconstitutedbytextualandunstructureddocuments,wheregeographicalreferencesareusuallygivenbymeansofplacena
2、mes.Acommonproblemwithtextualinformationretrievalisrepresentedbypolysemouswords,thatis,wordscanhavemorethanonesense.Thisproblemispresentalsointhegeographicaldomain:placenamesmayrefertodifferentlocationsintheworld.Inthispaperweinvestigatetheuseofourwordsensedisambiguationte
3、chniqueinthegeographicaldomain,withtheaimofresolvingambiguousplacenames.OurtechniqueisbasedonWordNetconceptualdensity.DuetothelackofareferencecorpustaggedwithWordNetsenses,wecarriedouttheexperimentsoverasetof1,210placenamesextractedfromtheSemCorcorpusthatwenamedGeoSemCoran
4、dmadepubliclyavailable.Wecomparedourmethodwiththemost-frequentbaselineandtheenhanced-Leskmethod,whichpreviouslyhasnotbeentestedinlargecontexts.Theresultsshowthatabetterprecisioncanbeachievedbyusingasmallcontext(phraselevel),whereasagreatercoveragecanbeobtainedbyusinglargec
5、ontexts(documentlevel).Theproposedmethodshouldbetestedwithothercorpora,duetothefactthatourexperimentsevidencedtheexcessivebiastowardsthemost-frequentsenseoftheGeoSemCor.Keywords:Wordsensedisambiguation;Toponymresolution;Conceptualdensity;Spatialindexing1.IntroductionAgreat
6、portionoftheinformationcurrentlyavailableindigitalformatisconstitutedbytextualandunstructureddocuments.Thecontinuousgrowthofthiskindofinformationandtheincreasingnumberofusersthatcanaccessitconstituteachallengetothedevelopersofinformationretrieval(IR)systems.Oneofthemostcha
7、llengingproblemsistheambiguityofhumanlanguage.Whensearchingforspecifickeywords,itisdesirabletoeliminateoccurrencesindocumentswherethewordorwordsareusedinaninappropriatesense(IdeandV′eronis1998).Ambiguitycanbeofvarioustypes:propernamesmayidentifydifferentclassesofnamedentit
8、ies(forinstance,‘London’mayidentifythewriter‘JackLondon’oracityintheUK),ormaybeusedasanam
此文档下载收益归作者所有