公开(公告)号：KR101329350B1

公开(公告)日：2013-11-14

申请号：KR1020120064528

申请日：2012-06-15

Applicant: 한국과학기술원

Inventor： 이민중 ,  이정민 ,  박예진 ,  최현 ,  정진완

IPC: G06F17/00 ,  G06F17/10

CPC classification number: G06F17/10

Abstract: The present invention relates to a betweenness centrality updating method of a graph. The method comprises (a) a step of obtaining a minimum cycle (hereinafter called the MUC) from an updated graph; (b) a step of extracting the MUC including a peak (hereinafter called the updated peak) transmitted to an updated edge (hereinafter called the MUCU) and calculating betweenness centrality (hereinafter called the local betweenness centrality) due to the shortest paths (hereinafter called the inner path) within the MUCU; (c) a step of obtaining betweenness centrality (hereinafter called the outer path increment) due to the shortest paths (hereinafter called the outer path) of the graph which is not included within the MUCU; and (d) a step of calculating betweenness centrality (hereinafter called the global betweenness centrality) of the graph by adding the outer path increament to the local betweenness centrality.

Abstract translation: 本发明涉及图形的中间性中心更新方法。 该方法包括（a）从更新的图形获得最小周期（以下称为MUC）的步骤; （b）提取MUC的步骤，包括发送到更新边缘的峰值（以下称为更新的峰值）（以下称为MUCU），并计算由于最短路径而导致的中心性（以下称为局部中心性） 内部路径）在MUCU内; （c）由于不包含在MUCU内的图形的最短路径（以下称为外部路径）而获得间隔中心性（以下称为外部路径增量）的步骤; 和（d）通过加上与局部中心性之间的外在路径来计算图的中心性（以下称为全局中心性）之间的步骤。

公开(公告)号：KR102252671B1

公开(公告)日：2021-05-17

申请号：KR1020190069460

申请日：2019-06-12

Applicant: 한국과학기술원

Inventor： 차미영 ,  박예진 ,  박건우 ,  김재헌 ,  강승체

IPC: G06Q50/30 ,  G06N20/00

Abstract: 혐오표현댓글필터링장치는, 혐오표현단어들을저장하는혐오표현사전, 단어임베딩모델을이용하여, 수집한댓글들에포함된혐오표현후보단어들과상기혐오표현단어들각각의임베딩벡터들을생성하고, 벡터간 유사도판단알고리즘을이용하여상기혐오표현후보단어들중에서상기혐오표현사전에저장된단어들과유사한유사단어들을추출하고, 상기유사단어들중 결정된신규혐오표현단어를상기혐오표현사전에저장하는사전구축장치, 상기혐오표현사전을기초로혐오표현댓글과비혐오표현댓글로라벨링된분류학습용댓글들을이용하여, 혐오표현분류모델이입력댓글을혐오표현댓글또는비혐오표현댓글을구별하도록학습시키는학습장치, 그리고판별대상댓글을상기혐오표현분류모델로입력하고, 상기혐오표현분류모델로부터상기판별대상댓글이혐오표현댓글로판별된경우, 상기판별대상댓글을미리설정된텍스트로치환되도록설정하는필터링기를포함한다.