• The KIPS Transactions:PartB
• /
• v.10B no.3
• /
• pp.347-358
• /
• 2003

VSField-associated term is a single or compound word whose terms occur in any document, and which makes it possible to recognize a field of text by using common knowledge of human. For example, human recognizes the field of document such as or , a field name of text, when she encounters a word 'Pitcher' or 'election', respectively We Proposes an efficient construction method of field-associated terms (FTs) for specializing field to decide a field of text. We could fix document classification scheme from well-classified document database or corpus. Considering focus field we discuss levels and stability ranks of field-associated terms. To construct a balanced FT collection, we construct a single FTs. From the collections we could automatically construct FT's levels, and stability ranks. We propose a new extraction algorithms of FT's for document classification by using FT's concentration rate, its occurrence frequencies.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.10d
• /
• pp.677-680
• /
• 2006

이 논문은 무선 센서 네트워크에서 최소 파워 구성 방법을 이용하여 에너지를 효율적으로 사용하면서 네트워크의 수명을 늘리고 지연(delay)을 줄이는 방안을 제시한다. 지금까지 에너지를 효율적으로 사용하기 위해 topology control, power-aware routing, sleep management 방법을 각각 독립적으로 연구되었는데 최소파워 구성은 위의 3가지 방법을 통합 최적화하여 에너지 효율적 사용 문제에 접근한다. 이러한 최소파워 구성 방법을 이용하여 선택적 웨이트 컨트롤(Selective weight control) 알고리즘을 제안하여 네트워크 전체 수명을 연장시키고 지연을 줄일 수 있다. 이 알고리즘은 각 노드가 주기적으로 잔류에너지를 인접노드들과 공유하여 에너지 랭크(rank)값을 유지한 상태에서 최소 에너지 경로를 사용하면서 에너지가 주위 노드들보다 작아질 때는 에너지가 많은 노드가 경로에 포함이 되도록 하여 특정 노드가 에너지 고갈되지 않고 네트워크내의 각 노드가 고르게 에너지를 소모하게 한다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2011.06a
• /
• pp.369-371
• /
• 2011

본 연구에서는 특허문서에 사용된 핵심키워드 찾아내고 추출된 핵심키워드에 가중치를 부여하여 특허데이터DB에서 질의문서와 유사한 특허기술문서를 찾고 유사도 순으로 우선 배치하여 검색에 효율을 높일 수 있는 알고리즘을 제안한다. 본 연구는 제안한 알고리즘은 검색결과에 대하여 질의한 문서와 유사한 문서 순으로 랭크가 가능하기 때문에 검색의 효율을 높이는 효과를 가지고 온다.

• Proceedings of the Korean Society for Information Management Conference
• /
• 2010.08a
• /
• pp.15-20
• /
• 2010

최근 Web of Science에 도입된 Eigenfactor지수와 논문 영향력 지수(Article Influence Score), 그리고 Scopus에 도입된 SJR 지수는 구글의 PageRank 알고리즘과 같은 네트워크 분석 방식의 인용지수이다. 국내 인용 색인 데이터베이스는 인용 링크가 외부로 향하는 비율과 자기 인용 비율이 높으므로 기존의 네트워크 인용 지수 산출 방식을 그대로 적용하기에는 어려움이 많다. 이 연구에서는 국내 인용색인DB에 대해서 대표적인 네트워크 인용 지수인 저널 페이지랭크를 시험적으로 측정해보고 국내 학술지의 상황을 고려한 개선방안을 모색하였다.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.15 no.5
• /
• pp.343-348
• /
• 2022

The STAP system suppresses clutter and jamming of the radar signal, but required a large number of samples for optimal performance. A large number of samples increases the signal processing computation. Therefore, there is need for a transformation method for reducing the signal rank. The LCMV beamforming method can easily set the distortion-free-constraint in the direction of arrival, and the beamforming scaling is excellent, so that overall rank can be reduced. In this study, the information of target is estimated using the proposed quadratic pattern constraints(QPC) and LCMV beamforming methods. The proposed method can perform beam pattern control in a desired direction according to the number of constraint conditions as a secondary pattern constraint condition. Through simulation, the performance of the propose method is verified. As a result on th simulation, the desired target was estimated when the proposed method had an angular resolution of 10 degrees or more, but it was not possible to accurately estimate the desired target when the angular resolution was less than 10 degrees.

• Journal of KIISE
• /
• v.44 no.2
• /
• pp.179-185
• /
• 2017

Document summarization aims to generate a summary that is consistent and contains the highly related sentences in a document. In this study, we implemented for document summarization that extracts highly related sentences from a whole document by considering both similarities and entailment relations between sentences. Accordingly, we proposed a new algorithm, TextRank-NLI, which combines a Recurrent Neural Network based Natural Language Inference model and a Graph-based ranking algorithm used in single document extraction-based summarization task. In order to evaluate the performance of the new algorithm, we conducted experiments using the same datasets as used in TextRank algorithm. The results indicated that TextRank-NLI showed 2.3% improvement in performance, as compared to TextRank.

• The Transactions of the Korea Information Processing Society
• /
• v.1 no.2
• /
• pp.184-193
• /
• 1994

This paper considers the Updating Minimum-weight Spanning Tree Problem(UMP), that is, the problem to update the Minimum-weight Spanning Tree(MST) in response to topology change of the network. This paper proposes the algorithm which reconstructs the MST after several links deleted and added. Its message complexity and its ideal-time complexity are Ο(m＋n log(t＋f)) and Ο(n＋n log(t＋f)) respectively, where n is the number of processors in the network, t(resp.f) is the number of added links (resp. the number of deleted links of the old MST), And m=t＋n if f=Ο, m=e (i.e. the number of links in the network after the topology change) otherwise. Moreover the last part of this paper touches in the algorithm which deals with deletion and addition of processors as well as links.

• Journal of Information Management
• /
• v.37 no.2
• /
• pp.33-50
• /
• 2006

In this paper, after reviewing hyperlink based ranking methods, we saw various other parameters that effect ranking. Then, We analyzed the PageRank and HITS(Hypertext Induced Topic Search) algorithm, which are two popular methods that use eigenvector computations to rank results in terms of their characteristics. Finally, google and Ask.com search engines were examined as examples for applying those methods. The results showed that use of Hyperlink structure can be useful for efficiency of web site search.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2009.04a
• /
• pp.383-386
• /
• 2009

집단지성을 이용한 지식검색 서비스는 개방적 구조와, 축적된 자료를 공유할 수 있다는 커뮤니티적인 특성으로 큰 인기를 얻고 있다. 하지만 방대한 지식공유속에서 사용자가 진정으로 원하는 답변 획득은 점점 더 어려워지고 있다. 최근 알고리즘적으로 가장 정교하다고 평가 받는 구글을 통해 상위에 랭크된 검색결과들 중에는 집단지성을 통해 구축된 위키피디아, Yahoo Q/A 과 같은 Social 검색엔진의 검색결과들이 상당수 존재한다. 본 논문은 대부분의 질문은 인간으로부터 문제해결의 실마리를 얻을 수 있다는 점과 온라인상의 사용자에 대한 연구를 통해 지식검색 서비스 사용자중 Influence를 찾는것에 목적이 있다. 이에 국내 Social 검색 엔진의 대표인 네이버 지식iN을 중심으로 지식검색내의 사용자 활동성과 신뢰성을 분석하고, 이를 기반으로한 Influence 지수 산정 알고리즘을 제안한다. 제안된 알고리즘을 통한 Influence 지수는 지식검색 서비스에서 문제 해결의 실마리를 가진 사용자를 찾는 중요한 지표가 될 것이다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.38B no.5
• /
• pp.385-393
• /
• 2013

In the past decade, a paradigm shift from machine-centered to human-centered and from technology-driven to user-driven has been witnessed. Consequently, Social search is getting more social and Social Network Service (SNS) is a popular Web service to connect and/or find friends, and the tendency of users interests often affects his/her who have similar interests. If we can track users' preferences in certain boundaries in terms of Web search and/or knowledge sharing, we can find more relevant information for users. In this paper, we propose a novel Topic Sensitive_Social Relationship Rank (TS_SRR) algorithm. We propose enhanced Web searching idea by finding similar and credible users in a Social Network incorporating social information in Web search. The Social Relation Rank between users are Social Relation Value, that is, for a different topics, a different subset of the above attributes is used to measure the Social Relation Rank. We observe that a user has a certain common interest with his/her credible friends in a Social Network, then focus on the problem of identifying users who have similar interests and high credibility, and sharing their search experiences. Thus, the proposed algorithm can make social search improve one step forward.