置顶 Python天然语言处理学习笔记 71doc.docx

置顶 Python天然语言处理学习笔记 71doc.docx

《置顶 Python天然语言处理学习笔记 71doc.docx》由会员分享，可在线阅读，更多相关《置顶 Python天然语言处理学习笔记 71doc.docx（7页珍藏版）》请在冰豆网上搜索。

置顶Python天然语言处理学习笔记71doc

置顶Python天然语言处理学习笔记71

网易博客保险提示：

体系检测到你当前密码的保险性较低，为了你的账号保险，倡议你适时修改密码破即修改|封闭

牛皮糖的旅程

成功的人并不是必定比别人更有才智，他们只是在始终的尝试、始终的失败，直至成功。

Python天然语言处理学习笔记（7）：

1.5主动懂得天然语言

2011-07-0200：

02：

11|分类：

PyNLTK|标签：

|字号大中小订阅原创翻译，如需转载，请与博主接洽：

新手上路，翻译不恰之处，恳请指出，不胜感激

1.5AutomaticNaturalLanguageUnderstanding主动懂得天然语言

Wehavebeenexploringlanguagebottom-up（自底向上的）,withthehelpoftextsandthePythonprogramminglanguage.However,we'realsointerestedinexploiting（利用）ourknowledgeoflanguageandcomputationbybuildingusefullanguagetechnologies.We'lltaketheopportunitynowtostepback（撤退）fromthenitty-gritty（细节）ofcodeinordertopaintabiggerpictureofnaturallanguageprocessing.

Atapurelypracticallevel,weallneedhelptonavigatetheuniverseofinformationlockedupintextontheWeb.SearchengineshavebeencrucialtothegrowthandpopularityoftheWeb,buthavesomeshortcomings（毛病）.Ittakesskill,knowledge,andsomeluck,toextractanswerstosuchquestionsas：

WhattouristsitescanIvisitbetweenPhiladelphia（费城）andPittsburgh（匹兹堡）onalimitedbudget?

WhatdoexpertssayaboutdigitalSLRcameras?

Whatpredictionsaboutthesteelmarketweremadebycredible（可托的）commentatorsinthepastweek?

Gettingacomputertoanswerthemautomaticallyinvolvesarangeoflanguageprocessingtasks,includinginformationextraction,inference（推理）,andsummarization（概括）,andwouldneedtobecarriedoutonascaleandwithalevelofrobustness（结实性）thatisstillbeyondourcurrentcapabilities（目前仍在咱们的才能之外）.

Onamorephilosophicallevel,along-standing（存在已经的）challengewithinartificialintelligencehasbeentobuildintelligentmachines,andamajorpartofintelligentbehaviorisunderstandinglanguage（智能行动的重要部分是语言的懂得）.Formanyyearsthisgoalhasbeenseenastoodifficult.However,asNLPtechnologiesbecomemoremature,androbustmethodsforanalyzingunrestrictedtextbecomemorewidespread,theprospect（期看）ofnaturallanguageunderstandinghasre-emerged（再次浮现）asaplausible（仿佛可托的）goal.

Inthissectionwedescribesomelanguageunderstandingtechnologies,togiveyouasenseoftheinterestingchallengesthatarewaitingforyou.

WordSenseDisambiguation词义消歧

Inwordsensedisambiguationwewanttoworkoutwhichsenseofawordwasintendedinagivencontext.Considertheambiguouswordsserveanddish：

（2）a.serve：

helpwithfoodordrink；holdanoffice；putballintoplayb.dish：

plate；courseofameal；communicationsdeviceInasentencecontainingthephrase：

heservedthedish,youcandetectthatbothserveanddisharebeingusedwiththeirfoodmeanings.It'sunlikelythatthetopicofdiscussionshiftedfromsportstocrockery（陶器）inthespaceofthreewords.Thiswouldforceyoutoinventbizarre（奇怪的）images,likeatennispro（职业网球）takingouthisfrustrationson（拿…出气）achinatea-setlaidout（摆设）besidethecourt（网球场）{这兴许会使得你面前产生一副怪异的画面，北京法律翻译公司一个职业网球手正把他的愁闷发泄到放在网球场边上的中国茶具}.Inotherwords,weautomaticallydisambiguatewordsusingcontext,exploitingthe*factthatnearbywordshavecloselyrelatedmeanings.Asanotherexampleofthiscontextual（高低文的）effect,considerthewordby,whichhasseveralmeanings,forexample,thebookbyChesterton（agentive（表示动作主体的词缀（如defendant的后缀-ant））-Chestertonwastheauthorofthebook）；thecupbythestove炉子（locative-thestoveiswherethecupis）；andsubmitbyFriday（temporal-Fridayisthetimeofthesubmitting）.Observein（3）thatthemeaningoftheitalicized（斜体的）wordhelpsusinterpretthemeaningofby.

（3）a.Thelostchildrenwerefoundbythesearchers（agentive）动词主体

b.Thelostchildrenwerefoundbythemountain（locative）地位

c.Thelostchildrenwerefoundbytheafternoon（temporal）时态

PronounResolution代词参较

Adeeperkindoflanguageunderstandingistoworkout"whodidwhattowhom,"i.e.,todetectthesubjects（主语）andobjects（宾语）ofverbs（动词）.Youlearnedtodothisinelementaryschool（小学）,butit'sharderthanyoumightthink.Inthesentencethethievesstolethepaintings,itiseasytotellwhoperformedthestealingaction.Considerthreepossiblefollowingsentencesin（4）,andtrytodeterminewhatwassold,caught,andfound（onecaseisambiguous）.

（4）a.Thethievesstolethepaintings.Theyweresubsequently（随后）sold.

b.Thethievesstolethepaintings.Theyweresubsequentlycaught.

c.Thethievesstolethepaintings.Theyweresubsequentlyfound.

Answeringthisquestioninvolvesfindingtheantecedent（先行词）ofthepronounthey,eitherthievesorpaintings.Computationaltechniquesfortacklingthisproblemincludeanaphoraresolution（指代消解）-identifyingwhatapronounornounphraserefersto-andsemanticrolelabeling（语义角色标注）-identifyinghowanounphraserelatestotheverb（asagent,patient,instrument,andsoon）.

antecedent：

【语法学】先行词（或语、led照明hardnesstester句）,前在词（或语、句）（如"IsawHarryandspoketohim"中的Harry对him而言）

anaphoraresolution：

【语法学】照顾，指代（法）（指用代词、助动词等调换某种语法成分的方法照顾或指代前面的词或词组，如Iknowitandhedoestoo.一句中的it跟do的利用）

semanticrolelabeling：

语义角色标注（SemanticRoleLabeling）指的是