-
公开(公告)号：KR1020140120711A
公开(公告)日：2014-10-14
申请号：KR1020130036933
申请日：2013-04-04
Applicant: 삼성전자주식회사
Inventor： 박형민
CPC classification number: G06F17/30539 , G06F17/30445 , G06F17/30483 , G06F2216/03
Abstract: 병렬 처리를 이용한 닫힌 빈발 패턴 마이닝 방법 및 장치에 관한 것으로, 일 양상에 따른 닫힌 빈발 패턴 마이닝 방법은 데이터베이스에서 빈발 아이템을 접두 패턴으로 하는 투영 데이터를 생성하여 저장하는 단계, 상기 저장된 투영 데이터간의 우선 순위에 따라 상기 저장된 투영 데이터를 복수의 프로세싱 노드에 할당하는 단계, 상기 할당된 투영 데이터를 이용하여 빈발 패턴을 생성하는 단계, 상기 빈발 패턴을 닫힌 빈발 패턴 리스트와 비교하여 상기 닫힌 빈발 패턴 리스트를 갱신하는 단계 및 상기 빈발 패턴이 상기 닫힌 빈발 패턴 리스트에 추가된 경우, 상기 빈발 패턴을 접두 패턴으로 하는 투영 데이터를 생성하여 저장하는 단계를 포함할 수 있다.
Abstract translation: 本发明涉及使用并行处理挖掘闭合频繁模式的方法和装置。 根据本发明的一个方面，一种挖掘闭合频繁模式的方法包括以下步骤：使用频繁项作为数据库中的前缀模式生成和存储投影数据; 根据存储的投影数据的优先级将存储的投影数据分配给多个处理节点; 使用所分配的投影数据产生频繁模式; 将频繁模式与闭合频繁模式列表进行比较，以更新关闭的频繁模式列表; 以及当频繁模式被添加到关闭的频繁模式列表时，使用频繁模式作为前缀模式生成和存储投影数据。
-
公开(公告)号：KR1020160066927A
公开(公告)日：2016-06-13
申请号：KR1020140172348
申请日：2014-12-03
Applicant: 삼성전자주식회사
IPC: A61B8/08
CPC classification number: G16H40/63 , A61B5/06 , A61B6/12 , A61B8/0833 , A61B8/5223 , G06F19/00 , G06T1/0078 , G16H50/20 , G16H50/30 , A61B8/08 , A61B8/0808 , A61B8/0816
Abstract: 컴퓨터보조진단지원장치및 방법이개시된다. 일실시예에따른컴퓨터보조진단지원장치는의료영상에대한진단결과의신뢰도가일정기준이하인경우신뢰도저조원인을분석하는원인분석부와, 분석된신뢰도저조원인에따라프로브의조작과관련된가이드정보를생성하는가이드정보생성부를포함할수 있다.
Abstract translation: 公开了一种用于支持计算机辅助诊断的装置和方法。 该装置包括：原因分析单元，被配置为分析医疗图像的诊断结果的可靠性降低的原因低于预定水平; 以及引导信息生成单元，被配置为基于所分析的可靠性劣化的原因来生成所述探测器的操作的引导信息。
-
公开(公告)号：KR1020140110157A
公开(公告)日：2014-09-17
申请号：KR1020130022972
申请日：2013-03-04
Applicant: 삼성전자주식회사
CPC classification number: G06F19/22
Abstract: The present invention relates to a method and device for searching for a pattern in sequence data. The method may comprise a step of determining an interest pattern model including the length, allowed inconsistent value, and minimum frequency of an interest pattern; a step of obtaining the frequency of similar patterns having an inconsistent value with a child pattern, which exceeds the allowed inconsistent value, by using inconsistent value information of the similar patterns of a mother pattern; and a step of determining whether the frequency of the child pattern satisfies at least one of the minimum frequencies by using the mother pattern and the frequency of similar patterns which exceeds the allowed inconsistent value.
Abstract translation: 本发明涉及一种用于搜索序列数据中的模式的方法和装置。 该方法可以包括确定包括兴趣模式的长度，允许不一致值和最小频率的兴趣模式模型的步骤; 通过使用母图案的类似图案的不一致的值信息来获得具有超过允许的不一致值的子图案具有不一致值的类似图案的频率的步骤; 以及通过使用母模式和超过允许的不一致值的类似模式的频率来确定子模式的频率是否满足最小频率中的至少一个的步骤。
-
公开(公告)号：KR1020120115005A
公开(公告)日：2012-10-17
申请号：KR1020110032898
申请日：2011-04-08
Applicant: 삼성전자주식회사 , 서울대학교산학협력단
IPC: G06F17/30
CPC classification number: G06F17/30634 , G06F17/30451 , G06F17/30011 , G06F17/30112 , G06F17/30327 , G06F17/30486
Abstract: PURPOSE: An efficient query processing method and a device thereof are provided to improve search performance by using a partial character string having an unfixed length. CONSTITUTION: A UI(User Interface)(105) receives a query character string and outputs a document including the character string. A database(165) stores information about various documents. A processor(135) generates an effective character string set composed of partial character strings having different lengths through the query character string. The processor determines one of subsets of the effective character string set as a candidate set based on costs of the subsets for accessing the database. The processor searches the database for the document including the query character string by using the candidate set. [Reference numerals] (105) User Interface; (110) Character string set generating unit; (120) Candidate set determining unit; (130) Document searching unit; (140) Index DB; (150) Inverted index DB; (160) Document DB
Abstract translation: 目的：提供一种高效的查询处理方法及其设备，通过使用具有不固定长度的部分字符串来提高搜索性能。 构成：UI（用户界面）（105）接收查询字符串并输出包含字符串的文档。 数据库（165）存储关于各种文档的信息。 处理器（135）通过查询字符串生成由具有不同长度的部分字符串组成的有效字符串集合。 处理器基于用于访问数据库的子集的成本来确定设置为候选集合的有效字符串的子集中的一个。 处理器通过使用候选集搜索数据库中包含查询字符串的文档。 （附图标记）（105）用户界面; （110）字符串组生成单元; （120）候选集确定单位; （130）文件检索单位; （140）指数DB; （150）反指数DB; （160）文件DB
-
-
Patent Agency Ranking
公开(公告)号：KR1020170034227A
公开(公告)日：2017-03-28
申请号：KR1020150132626
申请日：2015-09-18
Applicant: 삼성전자주식회사
CPC classification number: G10L15/14 , G10L15/063 , G10L15/07 , G10L15/08 , G10L15/18 , G10L15/183 , G10L2015/0635 , G10L2015/226 , G10L2015/227 , G10L2015/228
Abstract: 음성인식장치가개시된다. 음성인식장치는음향모델을이용하여, 입력된사용자의음성에대한음향점수를계산하는음향점수계산부와, 변환파라미터를이용하여계산된음향점수를표준음성에대한음향점수로변환하는음향점수변환부및 변환된음향점수를해석하여사용자의음성신호에대한인식결과를출력하는해석부를포함할수 있다.
Abstract translation: 公开了一种语音识别装置。 转换，使用的声学模型的语音识别装置声学得分的转换，并用于计算所述输入的用户的声音的声音分数计算单元的声学得分，使用变换参数声学得分在标准语音的声学得分的计算 以及解释单元，用于解释转换后的声音分数并输出用户的语音信号的识别结果。
-
公开(公告)号：KR101615164B1
公开(公告)日：2016-04-26
申请号：KR1020090023910
申请日：2009-03-20
Applicant: 삼성전자주식회사 , 서울대학교산학협력단
CPC classification number: G06F17/30625
Abstract: 엔-그램기반의질의처리장치및 그방법에관한기술이다. 질의어에대한전체엔-그램중에서일부의엔-그램을이용하여질의처리를수행한다. 상기일부의엔-그램에대한포스팅리스트(posting list)를이용하여상기질의어가존재할수 있는문서의후보셋(candidate set)을추출한다.
-
公开(公告)号：KR1020140108017A
公开(公告)日：2014-09-05
申请号：KR1020130022421
申请日：2013-02-28
Applicant: 삼성전자주식회사
CPC classification number: G06F17/30985 , G06F19/24 , G06K9/62
Abstract: The present invention relates to a method and device of searching for a pattern in sequence data. The method may comprise a step of determining a plurality of interest pattern models including the length, allowed inconsistent value, and minimum frequency of an interest pattern; a step of obtaining the frequency of a candidate pattern which can be generated within the length of the interest pattern based on the allowed inconsistent value of the interest pattern models; and a step of determining whether the frequency of the candidate pattern satisfies at least one of the minimum frequencies of the interest pattern models.
Abstract translation: 本发明涉及一种在序列数据中搜索模式的方法和装置。 该方法可以包括确定包括兴趣模式的长度，允许不一致值和最小频率的多个兴趣模式模型的步骤; 基于所述兴趣模式模型的所允许的不一致值，获得可以在兴趣模式的长度内生成的候选模式的频率的步骤; 以及确定候选模式的频率是否满足兴趣模式模型的最小频率中的至少一个的步骤。
-
公开(公告)号：KR1020140089675A
公开(公告)日：2014-07-16
申请号：KR1020130001258
申请日：2013-01-04
Applicant: 삼성전자주식회사
Abstract: An image search server, an image search service providing terminal, a system, and a method are disclosed. The image search server of the present invention comprises: a search request reception unit for receiving an image search request signal including identity information of neighboring objects and a search request image from the terminal; a search range determination unit for determining a search range in an object image stored based on the received terminal neighboring object identity information; a similar image searching unit for searching a similar object image with the search request image within the determined search range; and a search result transmission unit for transmitting a search result to the terminal.
Abstract translation: 公开了图像搜索服务器，图像搜索服务提供终端，系统和方法。 本发明的图像搜索服务器包括：搜索请求接收单元，用于从终端接收包括相邻对象的身份信息和搜索请求图像的图像搜索请求信号; 搜索范围确定单元，用于基于所接收的终端相邻的对象身份信息来确定存储的对象图像中的搜索范围; 类似的图像搜索单元，用于在确定的搜索范围内用搜索请求图像搜索类似的对象图像; 以及搜索结果发送单元，用于向终端发送搜索结果。
-
公开(公告)号：KR1020100105080A
公开(公告)日：2010-09-29
申请号：KR1020090023910
申请日：2009-03-20
Applicant: 삼성전자주식회사 , 서울대학교산학협력단
CPC classification number: G06F17/30625
Abstract: PURPOSE: A query processing method based on N-gram and an apparatus thereof are provided to improve query processing performance even when the length of a query is long. CONSTITUTION: A query processor(710) selects some N-grams among total N-grams for a query based on a query processing cost and uses a posting list for the some N-grams to extract a candidate set of a document in which a query exists. The query processor determines, based on the candidate set, the document including the query.
Abstract translation: 目的：提供一种基于N-gram及其装置的查询处理方法，以便即使查询长度较长，也可提高查询处理性能。 构成：查询处理器（710）基于查询处理成本为查询选择总N克中的一些N克，并使用一些N克的发布列表来提取其中查询的文档的候选集 存在。 查询处理器基于候选集确定包括查询的文档。
-
-
-
-
-
-
-
-
-
Search Results
27
records
(took 0.027 seconds)
