• 정보처리학회논문지D
• /
• 제9D권5호
• /
• pp.745-754
• /
• 2002

1998년에 등장한 구글 검색 사이트(http://www.google.com)에 처음 소개된 페이지랭크 알고리즘은 웹 문서들의 연결 구조에 기반하여 문서들간의 순위를 부여하는 방법이다. 페이지랭크 알고리즘은 상용 검색 엔진에서 구현되어 사용되고 있으나, 상업상의 이유들로 인하여 구현 기법에 관한 연구 결과는 거의 발표되지 않고 있다. [4,8]에서 소개된 페이지랭크 알고리즘의 구현 기법은 웹 문서들의 페이지랭크 값을 산출하기에 충분하지 않다. 본 논문은 페이지랭크 알고리즘의 구현 기법[4,8]을 설명하고, 이를 적용하는데 필요한 입/출력 자료 구조 및 4가지 주요 구현 기술을 제시한다. 본 논문은 실제 웹 문서의 페이지랭크 값을 산출하는 시스템을 예로 들어 페이지랭크 알고리즘을 적용하는 방법에 대한 이해를 돕도록 하였다.

• 한국정보전자통신기술학회논문지
• /
• 제11권5호
• /
• pp.467-474
• /
• 2018

다양한 업무 수행에 있어서 회의나 토론 등의 내용을 정리하여 문서화하는 것의 중요성은 매우 높다. 그러나 기존에는 사람이 직접 내용에 대한 정리를 수작업으로 수행하였다. 본 논문에서는 TextRank 알고리즘을 이용하여 자동으로 회의록을 생성하는 시스템의 개발에 대하여 설명한다. 제안한 시스템은 발언자의 모든 발언 내용을 실시간으로 기록하고, 문장들을 출현 빈도수에 기초하여 유사도를 계산한 후, 문서 데이터 안에서 문장들 간의 관계를 찾아내는 비지도 학습 알고리즘을 통해 중요 단어 혹은 문장을 추출함으로서 자동으로 회의록을 생성하도록 하였다. 특히, PageRank 알고리즘을 단어와 문장에 적합하도록 재구성한 TextRank 알고리즘에 대하여 핵심어의 가중치 조정 기법을 도입함으로서 성능 향상을 모색하였다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권1호
• /
• pp.109-134
• /
• 2018

Collaborative Filtering (CF) is widely used in recommendation field, which can be divided into rating-based CF and learning-to-rank based CF. Although many methods have been proposed based on these two kinds of CF, there still be room for improvement. Firstly, the data sparsity problem still remains a big challenge for CF algorithms. Secondly, the malicious rating given by some illegal users may affect the recommendation accuracy. Existing CF algorithms seldom took both of the two observations into consideration. In this paper, we propose a recommendation method based on listwise learning-to-rank by incorporating users' social information. By taking both ratings and order of items into consideration, the Plackett-Luce model is presented to find more accurate similar users. In order to alleviate the data sparsity problem, the improved matrix factorization model by integrating the influence of similar users is proposed to predict the rating. On the basis of exploring the trust relationship between users according to their social information, a listwise learning-to-rank algorithm is proposed to learn an optimal ranking model, which can output the recommendation list more consistent with the user preference. Comprehensive experiments conducted on two public real-world datasets show that our approach not only achieves high recommendation accuracy in relatively short runtime, but also is able to reduce the impact of malicious ratings.

• Journal of the Korean Statistical Society
• /
• 제26권3호
• /
• pp.333-354
• /
• 1997

Computational algorithms to calculate M-estimators and rank estimators of regression parameters from left-truncated and right-censored data are developed herein. In the case of M-estimators, new statistical methods are also introduced to incorporate leverage assements and concomitant scale estimation in the presence of left truncation and right censoring on the observed response. Furthermore, graphical methods to examine the residuals from these data are presented. Two real data sets are used for illustration.

• 대한수학회지
• /
• 제41권3호
• /
• pp.461-477
• /
• 2004

It is shown that every almost linear mapping h : A\longrightarrowB of a unital $C^{*}$ -algebra A to a unital $C^{*}$ -algebra B is a homomorphism under some condition on multiplication, and that every almost linear continuous mapping h : A\longrightarrowB of a unital $C^{*}$ -algebra A of real rank zero to a unital $C^{*}$ -algebra B is a homomorphism under some condition on multiplication. Furthermore, we are going to prove the generalized Hyers-Ulam-Rassias stability of *-homomorphisms between unital $C^{*}$ -algebras, and of C-linear *-derivations on unital $C^{*}$ -algebras./ -algebras.

• 응용통계연구
• /
• 제24권1호
• /
• pp.137-143
• /
• 2011

The classification of subjects with unknown distribution in a small sample size often involves order-restricted constraints in multivariate parameter setups. Those problems make the optimality of a conventional likelihood ratio based statistical inferences not feasible. Fortunately, Roy (1953) introduced union-intersection principle(UIP) which provides an alternative avenue. Multivariate linear rank statistics along with that principle, yield a considerably appropriate robust testing procedure. Furthermore, conditionally distribution-free test based upon exact permutation theory is used to generate p-values, even in a small sample. Applications of this method are illustrated in a real microarray data example (Lobenhofer et al., 2002).

• Communications for Statistical Applications and Methods
• /
• 제22권1호
• /
• pp.41-54
• /
• 2015

In this article, we propose nonconcave penalties on a reduced-rank regression model to select variables and estimate coefficients simultaneously. We apply HARD (hard thresholding) and SCAD (smoothly clipped absolute deviation) symmetric penalty functions with singularities at the origin, and bounded by a constant to reduce bias. In our simulation study and real data analysis, the new method is compared with an existing variable selection method using $L_1$ penalty that exhibits competitive performance in prediction and variable selection. Instead of using only one type of penalty function, we use two or three penalty functions simultaneously and take advantages of various types of penalty functions together to select relevant predictors and estimation to improve the overall performance of model fitting.

• International Journal of Aeronautical and Space Sciences
• /
• 제14권4호
• /
• pp.341-349
• /
• 2013

This paper considers the optimum design of flexbeam cross-sections for a full-scale bearingless helicopter rotor, using an efficient hybrid optimization algorithm based on particle swarm optimization, and an improved genetic algorithm, with an effective constraint handling scheme for constrained nonlinear optimization. The basic operators of the genetic algorithm, of crossover and mutation, are revisited, and a new rank-based multi-parent crossover operator is utilized. The rank-based crossover operator simultaneously enhances both the local, and the global exploration. The benchmark results demonstrate remarkable improvements, in terms of efficiency and robustness, as compared to other state-of-the-art algorithms. The developed algorithm is adopted for two baseline flexbeam section designs, and optimum cross-section configurations are obtained with less function evaluations, and less computation time.

• Journal of electromagnetic engineering and science
• /
• 제10권4호
• /
• pp.263-269
• /
• 2010

Multiple input multiple output (MIMO) has been considered one of the key enablers of broadband wireless communications. The indoor environment is known to be favorable to ensure both high rank property and high signal-to-interference/noise ratio (SINR) to fully exploit MIMO spatial multiplexing (SM) gain. In this paper, we describe several practical deployment cases where repeater links (or relay links), such as those present with an indoor distributed antenna system (DAS), can act as keyholes to degenerate the rank property of MIMO communications. In this case, we cannot exploit MIMO SM gain in indoor environment. We propose a novel MIMO communication scheme which uses simple converter in the devices in repeater links to resolve the rank degeneration issue and to ensure MIMO SM gain in highly MIMO-favorable indoor environment. MIMO SM is possible over the indoor DAS with single cable line through use of simple converters, which enables practical deployment in real fields.

• International Journal of Control, Automation, and Systems
• /
• 제3권3호
• /
• pp.430-443
• /
• 2005

A general construction algorithm of the Grassmann space parameters in linear systems - so-called, the Plucker matrix, 'L' in m-input, p-output, n-th order static output feedback systems and the Plucker matrix, $'L^{aug}'$ in augmented (m+d)-input, (p+d)-output, (n+d)-th order static output feedback systems - is presented for numerical checking of necessary conditions of complete static and complete minimum d-th order dynamic output feedback pole-assignments, respectively, and also for discernment of deterministic computation condition of their pole-assignable real solutions. Through the construction of L, it is shown that certain generically pole-assignable strictly proper mp > n system is actually none pole-assignable over any (real and complex) output feedbacks, by intrinsic rank deficiency of some submatrix of L. And it is also concretely illustrated that this none pole-assignable mp > n system by static output feedback can be arbitrary pole-assignable system via minimum d-th order dynamic output feedback, which is constructed by deterministic computation under full­rank of some submatrix of $L^{aug}$.