-
公开(公告)号：KR1020110057525A
公开(公告)日：2011-06-01
申请号：KR1020090113966
申请日：2009-11-24
Applicant: 한국전자통신연구원
IPC: G10L21/0272 , G10L15/20
Abstract: PURPOSE: A device for separating a sound source and a method thereof are provided to extract only a desired sound from various sound sources. CONSTITUTION: An input unit(610) changes the offered signal in to a frequency domain. A processing unit(620) divides the sound source of the converted signal in the frequency band unit. The processing unit aligns the separated sound source through the phase difference of a mixed filter for mixing the sound sources. An output unit(630) changes the aligned sound sources into the time domain.
Abstract translation: 目的：提供一种用于分离声源的装置及其方法，以从各种声源提取期望的声音。 构成：输入单元（610）将提供的信号改变为频域。 处理单元（620）以频带单位划分转换信号的声源。 处理单元通过用于混合声源的混合滤波器的相位差来对准分离的声源。 输出单元（630）将对准的声源改变为时域。
-
公开(公告)号：KR1020100072838A
公开(公告)日：2010-07-01
申请号：KR1020080131365
申请日：2008-12-22
Applicant: 한국전자통신연구원
CPC classification number: G10L15/08 , G10L15/142
Abstract: PURPOSE: A viterbi decoder and a method for recognizing a voice are provided to prevent the dramatic lowering of an observation probability of a contaminated portion caused by an unintended impulse noise. CONSTITUTION: An optimal state calculator(220) obtains the state of the maximum accumulated similarity in each measurement vector of an observation vector row for the inputted voice. A buffer unit(240) stores an observation probability value for the plural voices inputted prior to the inputted voice. A non-linear filtering unit(250) calculates the observation probability value based on the observation probability value calculated by an observation probability calculator(230). A maximum similarity producer(260) calculates a local maximum similarity value based on the observation probability value.
Abstract translation: 目的：提供维特比解码器和用于识别语音的方法，以防止由非预期脉冲噪声引起的污染部分的观察概率的显着降低。 构成：最佳状态计算器（220）获得输入语音的观测向量行的每个测量向量中的最大累积相似度的状态。 缓冲单元（240）存储在输入的语音之前输入的多个语音的观察概率值。 非线性滤波单元250基于由观测概率计算器计算出的观测概率值来计算观测概率值。 最大相似度生成器（260）基于观察概率值计算局部最大相似度值。
-
公开(公告)号：KR1020100072751A
公开(公告)日：2010-07-01
申请号：KR1020080131243
申请日：2008-12-22
Applicant: 한국전자통신연구원
IPC: G10L21/0208 , G10L15/20
Abstract: PURPOSE: A method and an apparatus for reducing noises are provided to reinforce isolation function of voice and noise through voice/noise isolation function like soft masking technique thereby accurately presuming clean voice. CONSTITUTION: A noise estimator(130) presumes noise component within inputted voice signal. A posterior probability estimator(140) presumes posterior probability value from the noise component. A noise parameter adapting unit(150) applies noise Gaussian mixture model to the inputted voice signal. A voice/noise separating unit(160) divides noise and voice signal primarily. A noise removing unit(170) eliminates residual noise components of the voice signal.
Abstract translation: 目的：提供减少噪声的方法和装置，通过语音/噪声隔离功能（如软掩蔽技术）来加强语音和噪声的隔离功能，从而准确地推断干净的声音。 构成：噪声估计器（130）假设输入的语音信号内的噪声分量。 后验概率估计器（140）假设来自噪声分量的后验概率值。 噪声参数适应单元（150）将噪声高斯混合模型应用于输入的语音信号。 语音/噪声分离单元（160）主要分离噪声和语音信号。 噪声去除单元（170）消除语音信号的残余噪声分量。
-
公开(公告)号：KR1020090063384A
公开(公告)日：2009-06-18
申请号：KR1020070130721
申请日：2007-12-14
Applicant: 한국전자통신연구원
CPC classification number: G06F17/279 , G06F17/2775
Abstract: A substitute reference solving apparatus performing national language analysis and a method thereof are provided to restore a substitute word of high accuracy in a complicated sentence and perform a national language analysis. A substitute recognizing unit(110) receives a character string and a substitute word is regarded. A semantic analyzing unit(120) analyzes that the whole meaning that substitute has. An entity name recognizing unit(130) recognizes clearly the name entity of the character string. A chunking unit allots a concept in a chunk by chunking and receiving the string. A substitute recovering unit compares the whole meaning and assigned concept of the substitute. The substitute decompression module selects the reference object used for the substitute reconstitution.
Abstract translation: 提供执行国家语言分析的替代参考解决装置及其方法，以复杂句子恢复高精度的替代词，并执行国家语言分析。 替代识别单元（110）接收字符串并考虑替代字。 语义分析单元（120）分析了替代物的全部含义。 实体名称识别单元（130）清楚地识别字符串的名称实体。 分块单元通过分块分配一个概念，并接收字符串。 替代回收单位比较替代品的整体含义和分配概念。 替代解压模块选择用于替代重建的参考对象。
-
公开(公告)号：KR1020090061520A
公开(公告)日：2009-06-16
申请号：KR1020070128550
申请日：2007-12-11
Applicant: 한국전자통신연구원
Abstract: A method for managing broadcast program names and person names in an EPG search service is provided to remove the necessity for a user to memorize and search a correct name included in an EPG DB. The original name is indexed through the space elimination of a program name included in an EPG DB(S200). A broadcast program name and a person name are separated by the syllable unit through an unused name list(S202). The space is removed from the program name of EPG and is indexed(S204). The program name is indexed by a bigram unit for robust recognition against a partial spelling error(S206).
Abstract translation: 提供了一种在EPG搜索服务中管理广播节目名称和人名的方法，以消除用户记住和搜索包括在EPG DB中的正确名称的必要性。 原始名称通过消除包含在EPG DB中的节目名称的空白来索引（S200）。 通过未使用的名称列表，音节单元将广播节目名称和人物名称分隔开（S202）。 将空间从EPG的节目名称中移除并被索引（S204）。 程序名称由一个二进制单位索引，用于强制识别部分拼写错误（S206）。
-
Patent Agency Ranking
公开(公告)号：KR1020090038187A
公开(公告)日：2009-04-20
申请号：KR1020070103554
申请日：2007-10-15
Applicant: 한국전자통신연구원
IPC: G06F17/40
CPC classification number: G06F17/30734
Abstract: An apparatus and a method for automatically generating ontology instance are provided to generate ontology instance from a document which does not exist in a knowledge server and a document which does not consist of the ontology form. A document collecting unit(100) collects a document corresponding to an ontology class. An unstructured document relation information extraction unit extracts relation information from the unstructured document. A structure document relation information extraction unit extracts relation information from the structure document. An instance generating unit(410) generates the ontology instance from the unstructured document relation information extraction unit and the relation information extracted from the structure document relation information extraction unit. The ontology instance generated in the instance generating unit is mapped to the corresponding class of the ontology by a mapping unit(420). In case the collected document is the unstructured document, the unstructured document is inputted the document collecting unit to the unstructured document relation information extraction unit. In case the collected document is the structure document, the structure document is inputted to the document collecting unit to the structure document relation information extraction unit.
Abstract translation: 提供了一种用于自动生成本体实例的装置和方法，用于从不存在于知识服务器中的文档和不包含本体表单的文档生成本体实例。 文档收集单元（100）收集对应于本体类的文档。 非结构化文档关系信息提取单元从非结构化文档中提取关联信息。 结构文件关系信息提取单元从结构文档中提取关系信息。 实例生成单元（410）从非结构化文档关系信息提取单元和从结构文档关系信息提取单元提取的关系信息生成本体实例。 在实例生成单元中生成的本体实例通过映射单元（420）映射到本体的对应类。 在收集的文档是非结构化文档的情况下，非结构化文档被输入到非结构化文档关系信息提取单元。 如果所收集的文档是结构文档，则结构文档被输入到文档收集单元到结构文档关系信息提取单元。
-
公开(公告)号：KR1020090030117A
公开(公告)日：2009-03-24
申请号：KR1020070095457
申请日：2007-09-19
Applicant: 한국전자통신연구원
CPC classification number: G06F17/279 , G10L15/063 , G10L15/1815 , G10L15/22
Abstract: An apparatus for post-processing conversation errors by using the multi-level verification in a voice conversation system and a method therefor are provided to recognize various conversation errors which can be generated in the conversation system, through the verification of multi-level type. A voice recognition part(50) extracts the feature vector of a voice signal and performs the voice recognition. A language analysis part(120) linguistically analyzes the user's utterance and outputs the language analysis result. A conversation analysis part(130) grasps the detailed meaning of the user's utterance based on the previous utterance and outputs the conversation analysis result. A conversation analysis and management part(140) analyzes the meaning of the user's utterance by referring to the flow of the whole conversation and outputs the analyzed result.
Abstract translation: 通过在语音对话系统中使用多级验证来处理会话错误的装置及其方法，用于通过多层次类型的验证来识别可在会话系统中产生的各种会话错误。 语音识别部分（50）提取语音信号的特征向量并执行语音识别。 语言分析部分（120）在语言上分析用户的话语并输出语言分析结果。 对话分析部（130）基于先前的发音来掌握用户的话语的详细含义，并输出会话分析结果。 对话分析和管理部分（140）通过参考整个会话的流程来分析用户话语的含义并输出分析结果。
-
公开(公告)号：KR1020080078147A
公开(公告)日：2008-08-27
申请号：KR1020070017809
申请日：2007-02-22
Applicant: 한국전자통신연구원
CPC classification number: G06F17/30914 , G06F17/3071
Abstract: A method and a device for extracting instances for constructing an ontology are provided to reduce time/expense needed for extracting the instances needed for constructing the ontology by extracting the instances automatically from a web document and apply the instances easily to constructing the ontology by converting the extracted instance into XML(eXtensible Markup Language). A class information storing part(100) stores a clue word classified into each class. A filter(102) removes an element not needed for extracting data by filtering a web document. A table processor(104) determines a row/column of a header by selecting a table for extracting the data from the filtered web document, classifying a type of the selected table, and referring to the clue word stored in the class information storing part. An instance processor(106) combines a head character string of the header row/column and the data corresponding to the header character string, and converts the extracted data into XML. The filter removes more than one of a tag representing image information, the tag for arranging text, the tag for display, and a cell used for layout without using data.
Abstract translation: 提供了一种用于提取用于构建本体的实例的方法和装置，以通过从web文档自动提取实例来减少提取构建本体所需的实例所需的时间/费用，并通过转换 提取的实例到XML（可扩展标记语言）。 类信息存储部（100）存储分类到各类的线索字。 过滤器（102）通过过滤web文档来去除提取数据所不需要的元素。 表处理器（104）通过选择用于从经过滤的网络文档中提取数据的表格，对所选择的表的类型进行分类，并参考存储在类信息存储部分中的线索字来确定报头的行/列。 实例处理器（106）组合标题行/列的头字符串和对应于标题字符串的数据，并将提取的数据转换成XML。 该过滤器除去代表图像信息的标签，用于排列文本的标签，用于显示的标签，以及用于布局的单元格，而不使用数据。
-
公开(公告)号：KR100750886B1
公开(公告)日：2007-08-22
申请号：KR1020050120977
申请日：2005-12-09
Applicant: 한국전자통신연구원
CPC classification number: G06F17/2818
Abstract: 본 발명은 정보검색, 정보추출, 번역, 자연어 처리 등의 작업을 위한 통계적 방법론에서 필요한 학습 데이터 구축을 효율적으로 하기 위한 학습 데이터 구축 장치 및 방법을 제공하기 위한 것으로서, (a) 학습 데이터에 대해 기계 학습을 수행하여 학습 모델을 생성하는 단계와, (b) 상기 생성된 학습 모델을 이용하여 원시 코퍼스에 자동으로 태그를 부착하여 학습 데이터 후보를 생성하는 단계와, (c) 상기 생성된 학습 데이터 후보의 신뢰점수를 계산하고, 계산된 후보의 신뢰 점수를 이용하여 학습 데이터 후보를 선택하는 단계와, (d) 사용자에게 인터페이스를 통해 상기 선택된 학습 데이터 후보에서 오류를 수정하고, 상기 오류 수정된 상기 학습 데이터 후보를 학습 데이터에 추가하여, 새로운 학습 모델을 점진적으로 확대시키는 단계를 포함하는데 있다.
학습데이터, 자동태그 부착, 학습 데이터 후보 선택, 능동 학습, 점진 학습-
公开(公告)号：KR1020070061080A
公开(公告)日：2007-06-13
申请号：KR1020060043312
申请日：2006-05-15
Applicant: 한국전자통신연구원
IPC: G06F17/40
CPC classification number: G06N5/02 , G06F17/2735 , G06F17/2795
Abstract: A knowledge normalization method for managing a knowledgebase and device thereof are provided to prevent performance degradation owing to disagreement between knowledge stored in the knowledgebase and a keyword used for searching the knowledge by normalizing the knowledge when an information extraction result is stored to the knowledgebase or the knowledge is searched from the knowledgebase. A memory part(210) stores normalization modes according to an attribute of an input character string and a normalization table(211) storing priority. A normalization controller(220) performs the normalization suitable for the attribute of the input character string based on information stored in the normalization table. A normalizer(230) normalizes the inputted character string by control of the normalization controller. The memory part includes a thesaurus(212), an abbreviation dictionary database(213), a sound different mark dictionary database(214), a pattern rule database(215), and a similar character string dictionary database(216). The normalizer includes a thesaurus-based normalizer(231), a dictionary-based normalizer(232), a rule-based normalizer(233), and a similar character string-based normalizer(234).
Abstract translation: 提供了一种用于管理知识库的知识规范化方法及其设备，用于防止存储在知识库中的知识之间的不一致以及当信息提取结果被存储到知识库时通过归一化知识来搜索知识的关键字的性能下降，或者 知识从知识库搜索。 存储器部件（210）根据输入字符串的属性和存储优先级的归一化表（211）来存储规范化模式。 规范化控制器220根据存储在标准化表中的信息执行适合于输入字符串的属性的归一化。 归一化器（230）通过归一化控制器的控制对输入的字符串进行归一化。 记忆部分包括辞典（212），缩写词典数据库（213），声音不同标记字典数据库（214），模式规则数据库（215）和类似的字符串字典数据库（216）。 归一化器包括基于词库的规范化器（231），基于字典的归一化器（232），基于规则的规范化器（233）和类似的基于字符串的归一化器（234）。
-
-
-
-
-
-
-
-
-
Search Results
128
records
(took 0.026 seconds)
