公开(公告)号：KR100341396B1

公开(公告)日：2002-06-22

申请号：KR1019990062419

申请日：1999-12-27

Applicant: 한국전자통신연구원

Inventor： 문병주 ,  정현수 ,  홍기채 ,  송종철 ,  이성용 ,  손소현

IPC: G06F17/20

Abstract: 본발명은기존의유사도를기반으로한 클러스터링(Clustering) 기법에계층단어(Hierarchy Terms) 개념을접목하여단어나문서의수평적연계뿐만아니라수직적연계, 즉디렉토리기반의단어및 문서분류까지가능하도록하는 3차원클러스터링생성시스템및 그방법을제공하는데그 목적이있다. 본발명에따르면, HTML(Hyper Text Markup Language) 및텍스트문서가입력되면, 상기 HTML 문서의실제내용을추출하고, 불용어를처리함으로써, 실제단어를추출한후, 상기추출된실제단어들에대한가중치(Weight)를부여하여색인파일을생성하는단어추출기; 입력된문서의단어들중 상기단어추출기에서부여된가중치가한계가중치(Threshold Weight, W)보다더 큰단어를추출하고, 상기추출된단어각각을중심어로하는단어군(클러스터)을생성하여, 미리저장된단어군과비교하여해당단어를중심으로하는새로운단어군을생성하는클러스터생성기; 및이용자의질의어가입력되면, 상기클러스터생성기에의하여생성된새로운단어군에서질의어에해당하는클러스터및 상기단어추출기에서생성한색인파일을이용하여문서값을출력하는 3-D 클러스터처리기를포함하여이루어진것을특징으로하는 3차원클러스터링생성시스템이제공된다.

公开(公告)号：KR1020020049164A

公开(公告)日：2002-06-26

申请号：KR1020000078266

申请日：2000-12-19

Applicant: 한국전자통신연구원

Inventor： 송종철 ,  문병주 ,  정현수 ,  홍기채 ,  손소현 ,  이성용

IPC: G06F17/21

CPC classification number: G06F17/3071

Abstract: PURPOSE: A system and a method for automatically classifying documents are provided to accurately obtain information on a document classified by categories, and to search and provide information in which the user is interested. CONSTITUTION: A morpheme analyzer(103) receives documents collected and link titles and extracts related terms. A term clustering generator(101) receives the terms extracted from the morpheme analyzer(103) and extracts keywords by documents. In addition, the term clustering generator(101) generates lists of the keywords by documents and term clusters. A gene learning classification device(102) receives the lists and the term clusters generated from the term cluster generator(101). In addition, the gene learning classification device(102) extracts term clusters for the keywords and infers categories of related fields.

Abstract translation: 目的：提供用于自动分类文档的系统和方法，以准确地获取按类别分类的文档的信息，并搜索和提供用户感兴趣的信息。 构成：语素分析器（103）接收收集的文件，并链接标题并提取相关术语。 术语聚类发生器（101）接收从语素分析器（103）提取的术语，并通过文档提取关键字。 此外，术语聚类生成器（101）通过文档和术语集合生成关键字的列表。 基因学习分类装置（102）接收从术语群发生器（101）生成的列表和项集合。 此外，基因学习分类装置（102）提取关键词的术语簇，并推断相关领域的类别。

公开(公告)号：KR100362380B1

公开(公告)日：2002-11-23

申请号：KR1019990062402

申请日：1999-12-27

Applicant: 한국전자통신연구원

Inventor： 정현수 ,  홍기채 ,  문병주 ,  김홍배 ,  송종철

IPC: G06F17/30

Abstract: 본발명은엑스엠엘(XML : eXtended Markup Language)을기반으로색인정보및 멀티미디어데이터를통합하여새로운형식의멀티미디어데이터를제작및 검색하는시스템과그를이용한멀티미디어데이터생성방법을제공하는데그 목적이있다. 본발명에따르면, 멀티미디어정보표현을위한 XML(eXtended Markup Language) 문서구조및 멀티미디어데이터를입력받아상기 XML 문서구조에의한멀티미디어정보를기술(Description)하여, 이를상기입력된멀티미디어데이터에삽입하여 XML 멀티미디어데이터로생성하는디스크립션제너레이터(Description Generator); 상기디스크립션제너레이터에의하여생성된 XML 멀티미디어데이터를멀티미디어정보및 검색정보로나누어색인하여저장한후, 검색요청메시지가수신되면, 검색을수행하여출력하는멀티미디어데이터검색서버; 및상기멀티미디어데이터검색서버에게검색요청메시지를전송하거나, 상기멀티미디어데이터검색서버로부터 XML 멀티미디어자료를입력받아전체또는씬(Scene) 단위로보여주는멀티미디어데이터검색플레이어를포함하여이루어진것을특징으로하는 XML 기반의멀티미디어데이터제작및 검색시스템이제공된다.

公开(公告)号：KR1020010064269A

公开(公告)日：2001-07-09

申请号：KR1019990062419

申请日：1999-12-27

Applicant: 한국전자통신연구원

Inventor： 문병주 ,  정현수 ,  홍기채 ,  송종철 ,  이성용 ,  손소현

IPC: G06F17/20

Abstract: PURPOSE: A system and method for creating a three dimensional clustering are provided to link a word to a document horizontally and to classify a word and a document based on a directory by grafting the hierarchy terms concept in a clustering method based on the existing similarity. CONSTITUTION: A word extracting device(101) extracts actual contents of an HTML document using an HTML DTD(Document type definition), and extracts an actual word out of filtered contents using an HTML filter and a dictionary(104), and creates an index file(107) by embodying a weight value by words. A cluster creating device(102) extracts a word having a value more than the critical weight value out of words of the corresponding document, and creates a word group(cluster) based on the extracted each word, and compares the word group(cluster) with the existing word group(cluster dictionary:105) based on the corresponding word, and creates new word group(cluster dictionary:105) based on the corresponding word. A 3-D cluster processor(103) checks whether a query language of a user exists in a classification dictionary(106). If a query language of a user does not exist in a classification dictionary, a document value is outputted using a cluster corresponded to the query language and the index file(107) in the cluster dictionary(105). If a query language of a user exists in a classification dictionary, the upper and lower linking word in the classification dictionary(106), a query language and a cluster with respect to the upper and lower linking word in the cluster dictionary(105) based on the upper and lower linking word are searched, and a document value is outputted using the cluster dictionary(105).

Abstract translation: 目的：提供一种用于创建三维聚类的系统和方法，用于将单词水平地链接到文档，并且基于目录通过基于现有相似度的聚类方法移植层次结构术语概念来对单词和文档进行分类。 构成：词提取装置（101）使用HTML DTD（文档类型定义）提取HTML文档的实际内容，并使用HTML过滤器和字典（104）从已过滤的内容中提取实际的单词，并创建索引 文件（107）通过用单词体现权重值。 集群创建装置（102）从对应文档的单词提取具有大于临界权值的值的单词，并且基于所提取的每个单词创建单词组（cluster），并将单词组（cluster） 与现有单词组（集群字典：105）基于相应的单词，并根据相应的单词创建新的单词组（集群字典：105）。 3-D集群处理器（103）检查分类字典（106）中是否存在用户的查询语言。 如果分类词典中不存在用户的查询语言，则使用与该集群字典（105）中的查询语言和索引文件（107）相对应的集群来输出文档值。 如果分类词典中存在用户的查询语言，则分类词典（106）中的上下连接词，查询语言和相对于集群辞典（105）中的上下连接词的集群 搜索上下连结词，并使用集群字典（105）输出文档值。

公开(公告)号：KR1020010064252A

公开(公告)日：2001-07-09

申请号：KR1019990062402

申请日：1999-12-27

Applicant: 한국전자통신연구원

Inventor： 정현수 ,  홍기채 ,  문병주 ,  김홍배 ,  송종철

IPC: G06F17/30

Abstract: PURPOSE: A system for creating and searching multi media data based on a XML and a method for creating multi media data using the system are provided to manufacture and search new-formed multi media data by unifying index information and multi media data based on a XML(extended markup language). CONSTITUTION: A description generator(110) receives a XML document structure for expressing multi media information and multi media data and describes multi media information by the XML document structure and inserts the information into the multi media data, and creates XML multi media data. A multi media data searching server(120) divides the XML multi media data into multi media information and search information and stores the information. The multi media data searching server(120) outputs information by performing a search if a search requesting message is received. A multi media data searching player(130) transmits the search requesting message to the multi media data searching server(120) and receives the XML multi media data from the multi media data searching server(120) and shows the data entirely or as the unit of a scene.

Abstract translation: 目的：提供一种基于XML创建和搜索多媒体数据的系统，以及使用该系统创建多媒体数据的方法，用于通过基于XML的索引信息和多媒体数据的统一来制作和搜索新形成的多媒体数据 （扩展标记语言）。 构成：描述生成器（110）接收用于表示多媒体信息和多媒体数据的XML文档结构，并通过XML文档结构描述多媒体信息，并将该信息插入到多媒体数据中，并创建XML多媒体数据。 多媒体数据搜索服务器（120）将XML多媒体数据划分为多媒体信息和搜索信息并存储信息。 如果接收到搜索请求消息，则多媒体数据搜索服务器（120）通过执行搜索来输出信息。 多媒体数据搜索播放器（130）将搜索请求消息发送到多媒体数据搜索服务器（120），并从多媒体数据搜索服务器（120）接收XML多媒体数据，并且完全显示数据或单元 的场景。