-
公开(公告)号:KR101735438B1
公开(公告)日:2017-05-15
申请号:KR1020100101419
申请日:2010-10-18
Applicant: 한국전자통신연구원
IPC: G06F17/28
Abstract: 실시간번역지식자동추출/검증방법및 그장치가개시된다. 실시간번역을위한번역지식자동추출/검증장치는실시간으로병렬말뭉치를수집하는실시간병렬말뭉치수집부와, 상기수집한병렬말뭉치에서미리정해진번역지식을추출하는번역지식추출부와, 상기추출한번역지식을이용하여제 1 평가요소와제 2평가요소로변환한후 검증하는번역지식검증부를포함하여구성될수 있다. 따라서, 웹에서매일생성되는병렬말뭉치를이용하여자동번역시스템의번역지식을실시간자동으로구축할수 있으므로기존자동번역지식사전을구축하는데소요되는구축비용을획기적으로줄일수 있는이점이있다.
Abstract translation: 公开了用于实时翻译知识的自动提取/验证的方法和装置。 使用实时平行语料库收集单元,用于收集平行语料库和翻译知识提取单元,所述提取的经翻译的知识,用于提取在收集hanbyeongryeol语料库预定用于实时翻译的翻译知识实时翻译知识自动提取/认证装置 翻译知识验证单元,用于将信息转换为第一评估因子和第二评估因子并对其进行验证。 因此,使用所产生的网络上每天都在平行文本,这样你就可以构建翻译知识转化为实时自动的机器翻译系统有一个点,可以降低显着楼宇自动化翻译词典的现有知识的建设成本。
-
公开(公告)号:KR101638442B1
公开(公告)日:2016-07-12
申请号:KR1020090113923
申请日:2009-11-24
Applicant: 한국전자통신연구원
Abstract: 구문규칙기반의구문분석에서구문분석의성능과정확도를높이기위해구문분절을수행하는것은영어, 한국어등 언어의구문분석모듈에서도사용되고있으나, 중국어는고립어로서굴절현상이없고, 단어사의에띄어쓰기를하지않아단어의경계가모호한특징을가지고있다. 대신, 중국어문장중에는문장기호를빈번하게사용하며, 중국어문장만이지니고있는문맥특징이있다. 본발명에서는, 중국어문장기호, 특정어휘, 문장의길이, 단어의품사등 중국어문맥정보를이용하여구문분절을수행함으로써, 구문규칙기반의중국어구문분석성능을높이고분석정확도를향상시킬수 있는중국어문맥정보를이용한중국어구문분절기술을제안하고자한다.
-
公开(公告)号:KR1020150113280A
公开(公告)日:2015-10-08
申请号:KR1020140036094
申请日:2014-03-27
Applicant: 한국전자통신연구원
IPC: G01C21/36 , G10L15/22 , G06F3/16 , G08G1/0969
CPC classification number: G01C21/3608 , G10L15/22 , G10L2015/223 , G06F3/16 , G08G1/0969
Abstract: 음성대화를이용한네비게이션제어장치및 방법이개시된다. 본발명에따른네비게이션제어장치는사용자로부터음성명령을획득하는음성획득부; 상기사용자의차량에설치되어있는네비게이션의구동정보및 상기차량의상태정보를포함하는차량데이터들을획득하는정보획득부; 상기음성명령을컴퓨터가이해할수 있는사용자의도로변환하는음성대화이해부; 상기차량데이터들중 상기사용자의도에상응하는관련정보를추출하는관련정보추출부; 및상기관련정보및 사용자의도를이용하여상기음성명령에대한대화응답을생성하고, 상기대화응답에상응하도록상기네비게이션을제어하는음성대화관리제어부를포함한다.
Abstract translation: 公开了一种使用语音对话来控制导航装置的装置和方法。 根据本发明的用于控制导航装置的装置包括从用户获取语音命令的语音获取单元; 信息获取单元获取包括关于车辆状态的信息和安装在用户车辆中的导航装置的操作的车辆数据; 语音对话理解单元将语音命令转换成计算机可以理解的用户命令; 信息抽取单元,从车辆数据抽取与用户命令对应的相关信息; 以及语音对话管理控制单元,通过使用相对信息和用户命令来生成关于语音命令的会话响应,并且控制导航以对应于会话响应。
-
公开(公告)号:KR1020150068663A
公开(公告)日:2015-06-22
申请号:KR1020130154590
申请日:2013-12-12
Applicant: 한국전자통신연구원
CPC classification number: G06F17/28 , G06F17/279
Abstract: 본발명은로그데이터를이용한자동번역장치및 그방법에관한것으로, 번역을요청한사용자로부터로그데이터를수집하는로그수집부; 상기사용자의단말로부터입력된발화문장및 상기로그데이터를분석한결과를토대로상기발화문장에해당하는발화상황을인식하고분류하는로그데이터처리부; 및분류된발화상황을토대로상기발화문장에대한번역지식을생성하고관리하는번역지식관리부;를포함하는것을특징으로한다.
Abstract translation: 本发明涉及使用日志数据的自动翻译装置及其方法,其中,所述装置包括:日志收集单元,用于从请求翻译的用户收集日志数据; 日志数据处理单元,用于基于分析语音发言语句的结果和由用户终端输入的日志数据来识别和分类对应于语音发言语句的情况; 以及翻译知识管理单元,用于基于分类的语音话语情况产生与语音话语句相关的翻译知识,并对其进行管理。
-
公开(公告)号:KR1020150035154A
公开(公告)日:2015-04-06
申请号:KR1020130115387
申请日:2013-09-27
Applicant: 한국전자통신연구원
CPC classification number: G06F17/30669 , G06F17/289 , G06F17/30696
Abstract: 자동번역된번역문에서선택된검증대상을이용하여검색정보를생성하고, 생성된검색정보로검색된검색결과들을정보검색소스의신뢰도및 정보검색결과의정확도에기반하여사용자에게제공하도록한 번역검증장치및 방법이제시된다. 제시된번역검증장치는원시언어문장을목적언어문장으로번역하여생성된최종목적언어번역문에포함된단어또는구문을검증대상으로선택하고, 선택된검증대상을근거로생성된복수의검색어들을근거로복수의정보검색소스들로부터검색결과들을검출하고, 정보검색소스의신뢰도및 정보검색결과의정확도를근거로설정된검색결과들의우선순위를근거로복수의검색결과들을제공한다.
Abstract translation: 提供了一种用于验证翻译的装置和方法,其使用从自动翻译文本中选择的验证目标生成搜索信息,并且基于信息搜索源的可靠性和信息搜索结果的准确性,通过生成的搜索信息搜索生成搜索结果。 所提供的用于验证翻译的装置选择以通过将用原始语言书写的句子翻译成以目标语言写成的语言而生成的目标语言中包含的最终翻译文本中包括的单词,短语或句子; 基于在所选择的验证对象的权限上生成的多个搜索词来检测来自多个信息搜索源的搜索结果; 并根据信息搜索源的可靠性和信息搜索结果的准确性,根据搜索结果集的优先级提供搜索结果。
-
公开(公告)号:KR1020140119841A
公开(公告)日:2014-10-13
申请号:KR1020130032522
申请日:2013-03-27
Applicant: 한국전자통신연구원
CPC classification number: G06T13/00 , G06F17/2854
Abstract: A method for verifying translation by using animation of the present invention can include a process of analyzing a first-language sentence of an original text inputted to be translated in a second language using a translation engine; a process of generating animation which can express a meaning of the first-language sentence using the analyzed result information of the first-language sentence; and a process of providing a user using the original text, with the original text and the generated animation to verify errors in translation.
Abstract translation: 通过使用本发明的动画来验证翻译的方法可以包括使用翻译引擎分析输入要以第二语言翻译的原始文本的第一语言句子的处理; 生成可以使用第一语言句子的分析结果信息来表达第一语言句子的意义的动画的过程; 以及使用原始文本提供用户原始文本和生成的动画以验证翻译中的错误的过程。
-
公开(公告)号:KR1020140079546A
公开(公告)日:2014-06-27
申请号:KR1020120146914
申请日:2012-12-14
Applicant: 한국전자통신연구원
Abstract: The present invention relates to a foreign language learning apparatus using a dialogue interface. The invention comprises: a corpus including large amount of English sentences; and the dialog interface for suggesting specific situations selectable by a user, for suggesting, to the user, missions to be pursued in the specific situation selected in order to enable the user to select a mission, and for performing conversations with the user. The invention is characterized by using the corpus to understand the intention of a spoken sentence of the user through the dialogue interface, secure link information between the intention understood and the spoken sentence of the user, examine if a grammatical error exists in the spoken sentence of the user, and provide the user with error-corrected knowledge along with an error examined by extracting the error-corrected knowledge corresponding to the examined error. The present invention like the same enables the user to experience various cases by dialogue scenarios of a system in a similar situation, and doubles learning effects by suggesting various expressions to the same user input sentence as a system response.
Abstract translation: 本发明涉及一种使用对话界面的外语学习装置。 本发明包括:包含大量英文句子的语料库; 以及用于建议用户可选择的特定情况的对话界面,用于向用户建议在所选择的特定情况下追求的任务,以使得用户能够选择任务,并且与用户进行对话。 本发明的特征在于使用语料库通过对话界面了解用户的口语句子的意图,确保用户意图理解和口语句子之间的安全链接信息,检查语音句子中是否存在语法错误 并且通过提取与所检查的错误相对应的错误校正的知识来向用户提供错误校正的知识以及检查的错误。 与此相同的本发明使得用户能够通过类似情况下的系统的对话场景来体验各种情况,并且通过向系统响应的相同用户输入语句建议各种表达来加倍学习效果。
-
公开(公告)号:KR1020140079545A
公开(公告)日:2014-06-27
申请号:KR1020120146912
申请日:2012-12-14
Applicant: 한국전자통신연구원
CPC classification number: G06F17/2755 , G06F17/10
Abstract: The present invention relates to a method for part-of-speech tagging and, more specifically, to a method for multi-language morphological analysis and part-of-speech tagging based on a phrase-based decoding mode. According to the configuration of the present invention, tagging ambiguity can be significantly reduced by taking phrases in part-of-speech tagging units without selecting words with high tagging ambiguity in tagging units and, additionally, various qualities can be effectively used by generalizing an existing word-based HMM model and proposing a robust probabilistic framework using a phrase-based decoding algorithm.
Abstract translation: 本发明涉及一种用于部分语音标签的方法,更具体地,涉及一种基于短语解码模式的多语言形态分析和部分语音标签的方法。 根据本发明的结构,可以通过在词性标签单元中使用短语而不在标签单元中选择具有高标签歧义的单词来显着地减少标记歧义,此外,可以通过将现有的 基于词的HMM模型,并提出了一种使用基于短语的解码算法的鲁棒概率框架。
-
公开(公告)号:KR1020130014730A
公开(公告)日:2013-02-12
申请号:KR1020110076361
申请日:2011-08-01
Applicant: 한국전자통신연구원
IPC: G06F17/28
CPC classification number: G06F17/2854 , G06F17/289
Abstract: PURPOSE: A user participation type translator tuning system is provided to update translation knowledge in real time by enabling a worker to access a translation tuning device through a network. CONSTITUTION: A translator(130) generates translation by receiving original sentences from a user terminal. A translation reliability evaluation device(140) evaluates the reliability of the translation by receiving the translation. A language resource management device(200) stores translation knowledge in a translation knowledge database(300) and includes a language resource collecting device which collects language resources in the Internet. A translation dictionary DB(DataBase)(310) is tuned based on the quality of the evaluated translation. A translation tuning device(100) includes a user authentication management device, a dictionary editor, the translator, the translation reliability evaluation device, a pub-web translator, a tuning sentence separator, and a tuning history tracking device. [Reference numerals] (110) User authentication management device; (120) Dictionary search editing device; (130) Translator; (140) Translation reliability evaluation device; (150) Manual evaluation device; (160) Public web translator; (170) Tuned sentence sorting device; (180) Tuned history tracking device; (210) Language resource searching device; (220) Language resource collecting device; (230) Web dictionary searching device; (240) Web searching device; (310) Translation dictionary DB; (320) Translated sentence buffer; (330) Tuned history DB; (340) Language resource DB; (40) Internet; (AA) Original text
Abstract translation: 目的:提供用户参与型翻译器调谐系统,通过使工作人员能够通过网络访问翻译调谐设备,实时更新翻译知识。 构成:翻译器(130)通过从用户终端接收原始句子来产生翻译。 翻译可靠性评估装置(140)通过接收翻译来评估翻译的可靠性。 语言资源管理装置(200)将翻译知识存储在翻译知识数据库(300)中,并且包括收集因特网中的语言资源的语言资源收集装置。 翻译字典DB(DataBase)(310)是根据评估翻译的质量进行调整的。 翻译调整装置(100)包括用户认证管理装置,字典编辑器,翻译器,翻译可靠性评估装置,公众网络翻译器,调音句子分离器和调谐历史跟踪装置。 (附图标记)(110)用户认证管理装置; (120)词典搜索编辑装置; (130)翻译; (140)翻译可靠性评估装置; (150)手动评估装置; (160)公共网页翻译; (170)调整句子分类装置; (180)调整历史追踪装置; (210)语言资源搜索装置; (220)语言资源收集装置; (230)网络词典搜索装置; (240)网页搜索装置; (310)翻译字典DB; (320)翻译句缓冲区; (330)调整历史DB; (340)语言资源DB; (40)互联网; (AA)原文
-
公开(公告)号:KR1020130014228A
公开(公告)日:2013-02-07
申请号:KR1020110076213
申请日:2011-07-29
Applicant: 한국전자통신연구원
IPC: G06F17/28
CPC classification number: G06F17/289 , G06F17/271 , G06F17/2755
Abstract: PURPOSE: An automatic translation method is provided to reduce errors in each step by creating a result based on the results created in a previous step. CONSTITUTION: A word class of morphemes in a source language is tagged. A source language syntactic tree is created according to the word class tagged in the morpheme(100). A purpose syntactic tree is created based on the source language syntactic tree(200). The purpose language is created based on the purpose language syntactic tree(400). An optimal purpose language is selected from the created purpose languages(500). After the appropriacy and fluency of the optimal purpose language are evaluated, the purpose language is selected based on an evaluation result. [Reference numerals] (100) Tagging at least one word class for each morpheme of a source language; (200) Generating at least one source language syntactic tree according to each word class tagged on the morpheme; (300) Generating purpose language syntactic tree based on each source language syntactic tree; (400) Generating at least one purpose language based on each purpose language syntactic tree; (510) Evaluating the appropriacy of the generated purpose language; (520) Evaluating the fluency of the generated purpose language; (530) Determining the ranking of generated purpose languages based on the appropriacy, the fluency, and the result of (400) step; (540) Selecting the optimal purpose language based on the determined ranking; (AA) Start; (BB) End
Abstract translation: 目的:提供一种自动翻译方法,通过基于上一步创建的结果创建一个结果来减少每个步骤中的错误。 构成:源语言中的词类词典被标记。 源语言句法树根据语素(100)中标记的单词类创建。 基于源语言句法树(200)创建目的语法树。 目的语言基于目的语言句法树(400)创建。 从创建的目的语言(500)中选择最佳用途语言。 在评估最佳目的语言的适当性和流畅性之后,基于评估结果选择目的语言。 (100)为源语言的每个语素标记至少一个单词类; (200)根据语素上标记的每个单词类生成至少一个源语言句法树; (300)基于每个源语言句法树生成目的语言句法树; (400)基于每个用途语言句法树生成至少一种用途语言; (510)评估所产生的目的语言的适用性; (520)评估所产生的目的语言的流畅程度; (530)根据(400)步骤的适当性,流畅性和结果确定生成的目的语言的排名; (540)基于确定的排名选择最优目的语言; (AA)开始; (BB)结束
-
-
-
-
-
-
-
-
-