• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.285-290
• /
• 2018

논지가 높은 설득력을 갖기 위해서는 충분한 지지 근거가 필요하다. 논지 내의 주장을 논리적으로 지지할 수 있는 근거 자료 추출의 자동화는 자동 토론 시스템, 정책 투표에 대한 의사 결정 보조 등 여러 어플리케이션의 개발 및 상용화를 위해 필수적으로 해결되어야 한다. 하지만 웹문서로부터 지지 근거를 추출하는 시스템을 위해서는 다음과 같은 두 가지 연구가 선행되어야 하고, 이는 높은 성능의 시스템 구현을 어렵게 한다: 1) 논지의 주제와 직접적인 관련성은 낮지만 지지 근거로 사용될 수 있는 정보를 확보하기 위한 넓은 검색 범위, 2) 수집한 정보 내에서 논지의 주장을 명확하게 지지할 수 있는 근거를 식별할 수 있는 인지 능력. 본 연구는 높은 정밀도와 확장 가능성을 가진 지지 근거 추출을 위해 다음과 같은 단계적 지지 근거 추출 시스템을 제안한다: 1) TF-IDF 유사도 기반 관련 문서 선별, 2) 의미적 유사도를 통한 지지 근거 1차 추출, 3) 신경망 분류기를 통한 지지 근거 2차 추출. 제안하는 시스템의 유효성을 검증하기 위해 사설 4008개 내의 주장에 대해 웹 상에 있는 845675개의 뉴스에서 지지 근거를 추출하는 실험을 수행하였다. 주장과 지지 근거를 주석한 정보에 대하여 성능 평가를 진행한 결과 본 연구에서 제안한 단계적 시스템은 1,2차 추출 과정에서 각각 0.41, 0.70의 정밀도를 보였다. 이후 시스템이 추출한 지지 근거를 분석하여, 논지에 대한 적절한 이해를 바탕으로 한 지지 근거 추출이 가능하다는 것을 확인하였다.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2008.05a
• /
• pp.1953-1956
• /
• 2008

본 연구에서는 금강의 하류부의 시간적 공간적 수질변화를 자세하게 분석하기 위하여 3차원 수리동역학 모델인 EFDC 와 비정상상태 수질모델인 WASP 7.2와 를 연계 적용하였다. 본 연구의 대상 구간은 금강하류 부중 대청댐 조절지 방류구 지점부터 공주의 정안천 유입 직전까지 총 48km의 구간이다. 하천의 수심별 수질은 일정하다고 가정하였으나 폭방향의 수질은 좌 중 우로 3개의 소구간으로 나누고 하천의 흐름 방향으로 나누어 전체적으로는 2차원적으로 구분하였다. 하천의 바닥 형상을 이용하여 모델 격자를 구성하였으며 수리모델은 건교부의 수위자료를 이용하여 보정하였고 수질자료는 환경부의 수질측정망 자료를 이용하여 보정하였다. 2차원 수리동역학-수질 연계모델링 결과, 대전 갑천이 유입되고 난 후의 청원지점에서 그리고 청주지역의 영향을 포함하는 미호천이 유입되고 난 연기 지점 그리고 본 연구의 최하류인 공주지점에서는 하천의 폭 방향으로 상당한 수질 차이가 있음을 확인하였다. 이는 대전 갑천 또는 청주 미호천 등이 유입된 이후 효과적으로 혼합되지 않는다는 것을 의미한다. 이는 해당지역의 수질시료 채취 위치에 따라 수질에 현격한 차이가 있을 수 있다는 것을 나타내며, 현재 환경부 수동측정망의 시료채취 방법을 고려하여 볼때 하천수질의 대표성을 나타낼 수 있는가 하는 문제의 차원에서 매우 심각하게 받아들여져야 할 것으로 생각된다. 한편, 현재 우리나라에서 시행되고 있는 오염총량관리제나 대부분의 환경영향평가에서는 하천의 폭 방향의 수질 차이에 대한 고려가 전무하며 이에 대한 전면적인 재검토가 시급하다고 판단된다.

• The Journal of Korean Association of Computer Education
• /
• v.18 no.4
• /
• pp.41-53
• /
• 2015

The purpose of this study is to investigate the effects of centralization on shared mental model, satisfaction and achievement in a web-based collaborative learning. Sixty-eight undergraduate students were randomly assigned to 7 groups and they participated in a 4-week web-based collaborative learning. Three highly centralized teams and three less centralized teams were selected, and analyzed for investigating whether there are meaningful differences in shared mental model, satisfaction and achievement according to the degree of centralization. The results showed that there was no significant difference in shared mental model between the highly and less centralized teams. However, the highly centralized teams showed higher level of satisfaction and achievement than the less centralized teams.

• The Journal of Korean Association of Computer Education
• /
• v.17 no.4
• /
• pp.69-78
• /
• 2014

This study aims to investigate better predictors, among learner participation and interaction, for collective self-efficacy and achievement in a web-based collaborative learning environment. Interaction requires communication among two or more learners, while participation does not. In this study, interaction was measured by in-degree centrality and out-degree centrality based on the social network analysis perspective. Multiple regression analysis results from 53 college students who performed team project via online showed that in-degree centrality predicted collective self-efficacy and out-degree centrality predicted achievement, while participation was not a significant predictor.

• The Transactions of the Korea Information Processing Society
• /
• v.2 no.2
• /
• pp.215-228
• /
• 1995

Despite their importance in language, there have been relatively few computational studies in understanding words. This paper describes how neural networks can learn to perceive and produce words. Most traditional linguistic theories presuppose abstract underlying representations (UR) and a set of explicit rules to obtain the surface realization. There are, however, a number of questions that can be raised regarding this approach: (1) assumption of URs, (2) formation of rules, and (3) interaction of rules. In this paper, it is hypothesized that rules would emerge as the generalizations the network abstracts in the process of learning to associate forms with meanings of the words. Employing a simple recurrent network, a series of simulations on different types of morphophonemic processes was run. The results of the simulations show that this network is capable of learning to perceive whether words are in basic from or in inflected form, given only forms, and to produce words in the right form, given arbitrary meanings, this eliminating the need for presupposing abstract URs and rules.

• Journal of Internet Computing and Services
• /
• v.13 no.5
• /
• pp.65-75
• /
• 2012

Organizational citizenship behaviors(OCB) are discretionary behaviors of employees, which go beyond that which is required, and are known to be contributing factors of organizational performance. When a supply chain management(SCM) system is implemented, organizational knowledge concerning the global standards, process innovation and key performance indicators(KPI) is also spread out across different, disconnected silos, thus increasing the absorptive and innovative capacity of employees. Therefore, OCB may be an important antecedent to successful operation of supply chain. This paper examines the causal relationships among absorptive capacity of employees, organizational citizenship, SCM performance and intention to innovate in global manufacturing corporations in Korea. Empirical results from 122 survey responses indicate that the organizational citizenship affects the level of SCM performance and absorptive capacity, which, in turn, influences SCM performance. As expected, SCM performance has been found to affect intention to innovate, but absorptive capacity has no impact on intention to innovate. As a conclusion, the academic and practical implications of these findings are discussed.

• Journal of Korean Society of Transportation
• /
• v.25 no.4
• /
• pp.99-108
• /
• 2007

This paper presents an elastic demand stochastic user equilibrium traffic assignment that could not be easily tackled. The elastic demand coupled with a travel performance function is known to converge to a supply-demand equilibrium, where a stochastic user equilibrium (SUE) is obtained. SUE is the state in which all equivalent path costs are equal, and thus no user can reduce his perceived travel cost. The elastic demand SUE traffic assignment can be formulated based on a dynamic system, which is a means of describing how one state develops into another state over the course of time. Traditionally it has been used for control engineering, but it is also useful for transportation problems in that it can describe time-variant traffic movements. Through the Lyapunov Function Theorem, the author proves that the model has a stable solution and confirms it with a numerical example.

• Journal of Internet Computing and Services
• /
• v.7 no.1
• /
• pp.75-89
• /
• 2006

This paper is concerned with analysing the bankruptcy prediction power of three models: Multivariate Discriminant Analysis(MDA), Logit Analysis, Neural Network. The research targeted the bankrupted companies after the foreign exchange crisis in 1997 to differentiate from previous research efforts, and all participating companies were randomly selected from the KSE listed companies belonging to manufacturing industry to improve prediction accuracy and validity of the model. In order to assure meaningful bankruptcy prediction, training data and testing data were not extracted within the corresponding period. The result is that prediction accuracy of neural networks is more excellent than that of logit analysis and MDA model when considering that execution of testing data was followed by execution of training data.

• Journal of Internet Computing and Services
• /
• v.20 no.6
• /
• pp.65-72
• /
• 2019

Data augmentation is a method that increases the amount of data through various algorithms based on a small amount of sample data. When machine learning and deep learning techniques are used to solve real-world problems, there is often a lack of data sets. The lack of data is at greater risk of underfitting and overfitting, in addition to the poor reflection of the characteristics of the set of data when learning a model. Thus, in this paper, through the layer-wise data augmenting method at each layer of deep neural network, the proposed method produces augmented data that is substantially meaningful and shows that the method presented by the paper through experimentation is effective in the learning of the model by measuring whether the method presented by the paper improves classification accuracy.

• Journal of Biomedical Engineering Research
• /
• v.24 no.3
• /
• pp.151-158
• /
• 2003

Numerous studies of short-term, beat-to-beat variability in cardiovascular signals have used linear analysis techniques. However, no study has been done about the appropriateness of linear techniques or the comparison between linearities and nonlinearities in short-term, beat-to-beat variability. This paper aims to verify the appropriateness of linear techniques by investigating nonlinearities in short-term, beat-to-beat variability. We compared linear autoregressive moving average(ARMA) with nonlinear neural network(NN) models for predicting current instantaneous heart rate(HR) and mean arterial blood pressure(BP) from past HRs and BPs. To evaluate these models. we used HR and BP time series from the MIMIC database. Experimental results indicate that NN-based nonlinearities do not play a significant role and suggest that 10 technique provides adequate characterization of the system dynamics responsible for generating short-term, beat-to-beat variability.