公开(公告)号：KR1020080068273A

公开(公告)日：2008-07-23

申请号：KR1020070005802

申请日：2007-01-18

Applicant: 한양대학교 산학협력단

Inventor： 김상욱 ,  장지웅 ,  배덕호

IPC: G06F12/00 ,  G06F12/02 ,  G06F12/06

Abstract: An apparatus and a method for managing a flash memory are provided to correct or delete a record directly in a flash memory by using a overwrite operation to reduce allocation of a new page, thereby improving the performance of the flash memory and decreasing the number of erase operations. An apparatus for managing a flash memory includes a data reading unit(210), a data processor(240), and a status information processor(250). The data reading unit reads a to-be-accessed page from the flash memory. The data processor selects a to-be-recorded slot in a first state corresponding to an initial state from slots which construct the to-be-accessed page and records the record of a to-be-corrected slot, which has been corrected. The status information processor corrects the status information of the to-be-recorded slot from the first state to a second state which represents that a record is stored, corrects the status information of the to-be-corrected slot from the second state to a third state which represents that a record is moved and records the address of the to-be-recorded slot in the address region of the to-be-corrected slot.

Abstract translation: 提供了一种用于管理闪速存储器的装置和方法，以通过使用重写操作直接在闪速存储器中校正或删除记录，以减少新页面的分配，从而提高闪存的性能并减少擦除次数 操作。 一种用于管理闪速存储器的装置，包括数据读取单元（210），数据处理器（240）和状态信息处理器（250）。 数据读取单元从闪速存储器读取要访问的页面。 数据处理器从构成待访问页面的时隙中选择对应于初始状态的第一状态的待记录时隙，并记录已被校正的待校正时隙的记录。 状态信息处理器将待记录时隙的状态信息从第一状态修改为表示存储记录的第二状态，将待校正时隙的状态信息从第二状态修正为 表示记录被移动的第三状态，并且将要被记录的时隙的地址记录在待校正的时隙的地址区域中。

公开(公告)号：KR101906859B1

公开(公告)日：2018-10-11

申请号：KR1020120030043

申请日：2012-03-23

Applicant: 삼성전자주식회사 ,  한양대학교 산학협력단

Inventor： 김상욱 ,  김상욱 ,  배덕호

IPC: G06F17/30

Abstract: 본발명은시계열데이터에서이상서브시퀀스를검출하는이상서브시퀀스검출장치및 방법에관한것이다. 본발명의일 실시예에의한이상서브시퀀스검출장치는시계열데이터를기호화하고, 상기기호화된문자열을이용해접미사트리를구축하는전처리부, 상기구축된접미사트리를순회방문하면서최대패턴의길이를갱신하고, 상기갱신된최대패턴의길이의최대값과최소값을기초로윈도우길이범위를결정하는윈도우길이결정부, 상기결정된윈도우길이별로서브시퀀스와상기서브시퀀스의 k번째이웃서브시퀀스와의격자거리가가장큰 top-k개의이상서브시퀀스후보를도출하는이상후보도출부, 상기도출된이상서브시퀀스후보에대해이상치점수를계산하는이상치계산부, 및상기계산된이상치점수를기초로 top-k개의이상서브시퀀스를검출하는이상서브시퀀스검출부를포함한다.

公开(公告)号：KR1020140120970A

公开(公告)日：2014-10-15

申请号：KR1020130036249

申请日：2013-04-03

Applicant: 한양대학교 산학협력단

Inventor： 김진형 ,  김상욱 ,  배덕호 ,  박찬익

IPC: G06F12/02

Abstract: 인텔리전트 SSD(Solid State Disk)의 구조적 특징을 통해 대용량의 데이터를 정렬하는 외부 정렬 알고리즘을 효율적으로 수행하는 방안이 개시된다. 그룹화를 이용한 외부 정렬 알고리즘은 빈번한 데이터 접근을 방지할 수 있고, 런의 수에 상관없이 데이터를 관리 대용량 데이터의 쓰기 작업을 최소화할 수 있어 외부 정렬 수행 시 여러 개의 런들을 줄임으로써 효율적인 외부 정렬을 수행할 수 있다.

Abstract translation: 公开了一种通过智能固态盘（SSD）的结构特征有效地执行排序大数据的外部排序算法的方法。 使用分组的外部排序算法可以防止频繁的数据访问，并且可以通过管理数据来最小化关于大数据的写入操作，而不管运行次数。 因此，本发明可以减少外部排序操作期间所需的运行次数，以实现高效的外部分选操作。

公开(公告)号：KR101399826B1

公开(公告)日：2014-05-27

申请号：KR1020120158298

申请日：2012-12-31

Applicant: 한양대학교 산학협력단

Inventor： 배덕호 ,  이종민 ,  강병호 ,  김태훈 ,  김상욱

IPC: G06F17/00 ,  G06Q10/06

CPC classification number: G06Q10/063112 ,  G06F17/10 ,  G06Q10/063118

Abstract: Disclosed are a system and a method for organizing a team to maximize a profit in a social network. The method that a team organizing system organizes a team to perform a project comprises a step of calculating the rarity of a skill required to perform the project among skills which members have and the rarity of the member; and a step of assigning non-rare members to the project included in a project tool by priority based on the calculated rarity of the skill and the member.

Abstract translation: 公开了一种用于组织团队以使社交网络中的利润最大化的系统和方法。 团队组织系统组织团队执行项目的方法包括计算成员所具有的技能和成员的稀缺性之间执行项目所需的技能的稀少性的步骤; 并且基于所计算的技能和成员的稀有度，优先将项目工具中包含的非罕见成员分配给项目工具。

公开(公告)号：KR1020130107889A

公开(公告)日：2013-10-02

申请号：KR1020120030043

申请日：2012-03-23

Applicant: 삼성전자주식회사 ,  한양대학교 산학협력단

Inventor： 김상욱 ,  김상욱 ,  배덕호

IPC: G06F17/30 ,  G06F19/00

CPC classification number: G06F17/30303

Abstract: PURPOSE: A device and a method for detecting abnormal subsequence are provided to detect the abnormal subsequence group with the same length and shape, thereby efficiently detecting the abnormal subsequence displayed in various domains. CONSTITUTION: A preprocessing unit (221) symbolizes time series data and uses the symbolized string, thereby building a suffix tree. A window length determining unit (222) updates the length of the maximum pattern with round-visiting the suffix tree, and determines the range of window length based on the maximum value and the minimum value of the maximum pattern length. The abnormal candidate deduction unit (223) deducts the abnormal sequence candidates which has the largest length difference between the subsequence and the neighbor subsequence for each window length. The abnormal value calculation unit (224) calculates an abnormal value with respect to the abnormal subsequence candidate, and the abnormal subsequence detecting unit (225) detects the abnormal subsequences based on the abnormal scores. [Reference numerals] (221) Preprocessing unit; (222) Window length determining unit; (223) Abnormal candidate deduction unit; (224) Abnormal value calculation unit; (225) Abnormal subsequence detecting unit

Abstract translation: 目的：提供一种用于检测异常子序列的装置和方法，用于检测具有相同长度和形状的异常子序列，从而有效地检测各个域中显示的异常子序列。 构成：预处理单元（221）表示时间序列数据并使用符号串，从而构建后缀树。 窗口长度确定单元（222）通过循环访问后缀树来更新最大模式的长度，并且基于最大模式长度的最大值和最小值来确定窗口长度的范围。 异常候选扣除单元（223）扣除每个窗口长度的子序列和相邻子序列之间具有最大长度差的异常序列候选。 异常值运算部（224）针对异常子序列候选计算异常值，异常子序列检测部225根据异常得分检测异常子序列。 （附图标记）（221）预处理单元; （222）窗口长度确定单元; （223）候选扣除单位异常; （224）异常值计算单位; （225）异常子序列检测单元