• The Journal of Korean Association of Computer Education
• /
• v.11 no.1
• /
• pp.65-74
• /
• 2008

In this paper, we introduce an analysis system for answer dataset by using a data mining method. We analyze students' answer data collected from a test including multiple choice question items, and find associations between the items. Analysis of evaluation results based on our system will not only provide correct information on students' achievement levels but also provides a basis for modifying weaknesses of the evaluation procedures, question items, or teaching/learning procedures. Furthermore, it will enable us to improve the quality of question items for future use so that we can secure itemsets of high quality.

• Annual Conference on Human and Language Technology
• /
• 1997.10a
• /
• pp.102-108
• /
• 1997

기존의 품사 태깅 방법에서 독립적으로 사용해온 언어 지식과 통계 정보는 품사 태깅의 정확도와 처리 범위의 향상을 위해서 상호 보완적인 특성을 갖는다. 이에 본 논문은 언어 지식과 통계 정보의 보완적 특성을 이용한 규칙 우선 직렬 품사 태깅 방법을 제안한다. 제안된 방법은 언어 지식에 의한 품사 태깅 결과를 선호함으로써 규칙 기반 품사 태깅의 정확도를 유지하며, 언어 지식에 의해서 모호성이 해소되지 않은 어절에 통계 정보에 의한 품사 태깅 결과를 할당함으로써 통계 기반 품사 태깅의 처리 범위를 유지한다. 또한, 수정 언어 지식에 의해 태깅 결과의 오류를 보정함으로써 품사 태깅의 정확도를 향상시킨다. 약 2만 어절 크기의 외부 평가 코퍼스에 대해 수행된 실험 결과, 규칙 우선 직렬 품사 태깅 시스템은 통계 정보만을 이용한 품사 태깅의 정확도보다 32.70% 향상된 95.43%의 정확도를 보였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2002.11c
• /
• pp.2227-2230
• /
• 2002

본 논문에서는 WWW을 이용한 전자상거래 환경(B2B, B2C, 등)에서 기업간(시스템 사이) 원활한 업무 협조와 just-in-time 서비스를 구현할 수 있도록, 능동 데이터베이스 기법을 이용한 전자상거래 시스템의 능동 kernel을 제안한다. 능동 커널은 고수준(high level)의 규칙 프로그래밍 언어(ECA rule)를 제공하므로 응용 프로그램과는 독립적으로 기업간 업무협조(coordination)와 사건 기반의 just-in-time 서비스를 구현할 수 있다. 따라서 시스템 관리자 및 프로그래머는 전자상거래 시스템간 조정 기능 구현과 유지보수를 쉽게 할 수 있다. 또한, 본 논문에서 제안하는 능동 커널은 기업의 방화벽(fire-wall)을 통해서도 적용될 수 있도록 HTTP 프로토콜을 이용하므로, platform 독립적인 구현을 통하여 기업간 시스템의 보안 유지와 독립성을 보장할 수 있다.

• The KIPS Transactions:PartB
• /
• v.11B no.2
• /
• pp.199-206
• /
• 2004

This paper proposes a hybrid independent component analysis(ICA) of fixed-point(FP) algorithm and robust algorithm. The FP algorithm is applied for improving the analysis speed and performance, and the robust algorithm is applied for preventing performance degradations by means of very small kurtosis and temporal correlations between components. And the adaptive adaptation of temporal correlations has been proposed for solving limits of the conventional robust algorithm dependent on the maximum time delay. The proposed ICA has been applied to the problems for separating the 4-mixed signals of 500 samples and 10-mixed images of $512\times512$pixels, respectively. The experimental results show that the proposed ICA has a characteristics of adaptively adapting the maximum time delay, and has a superior separation performances(speed, rate) to conventional FP-ICA and hybrid ICA of heuristic correlation. Especially, the proposed ICA gives the larger degree of improvement as the problem size increases.

• Smart Media Journal
• /
• v.5 no.1
• /
• pp.104-113
• /
• 2016

A new design method for multi-functional pull-down menu based on stacking rules is proposed. The importance of a design point of view has been emphasized in many previous studies on web navigation. Taking technical aspects with an emphasis on programming techniques into consideration in this paper, multiple elements, which are used for constructing main menu of the pull-down menu, are capable of performing their unique functions independently in addition to basic classification function by disposing them to separated layers using stacking rules. Furthermore, the improvement of visual effects and efficiency for information navigation can be expected by implementing submenu selection-bar using animation techniques and function to automatically search submenu item. Finally, the effectiveness of the proposed method is identified by implementing multi-functional pull-down menu using CSS and jQuery.

• The Korean Journal of Applied Statistics
• /
• v.21 no.4
• /
• pp.615-620
• /
• 2008

In engineering experiments, 'Sometimes Pooling Rules' to remove insignificant terms from the model has been implemented to increase the power of detecting the small size of main effects when the preliminary test of higher order interaction effects declare to be insignificant. In this note, we review the sometimes pooling rules in the literature and also study the probability of the length of 95% confidence interval of ${\mu}_1-{\mu}_2$ of the comparison of two independent samples being shorter than that of the paired comparison at the various level of significance ${\alpha}_1$ of the preliminary test and the insufficient number of blocks n in [2,13], given the block effects being pooled to the error term. This study supports that the sometimes pooling results in the power improvement of the main effects.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.13 no.2
• /
• pp.200-208
• /
• 2003

This paper proposes a separation and feature extraction of image signals using the independent component analysis(ICA) based on neural networks of efficient learning rule. The proposed learning rule is a hybrid fixed-point(FP) algorithm based on secant method and momentum. Secant method is applied to improve the performance by simplifying the 1st-order derivative computation for optimizing the objective function, which is to minimize the mutual informations of the independent components. The momentum is applied for high-speed convergence by restraining the oscillation in the process of converging to the optimal solution. The proposed algorithm has been applied to the composite images generated by random mixing matrix from the 10 images of $512\times512$-pixel. The simulation results show that the proposed algorithm has better performances of the separation speed and rate than those using the FP algorithm based on Newton and secant method. The proposed algorithm has been also applied to extract the features using a 3 set of 10,000 image patches from the 10 fingerprints of $256\times256$-pixel and the front and the rear paper money of $480\times225$-pixel, respectively, The simulation results show that the proposed algorithm has also better extraction speed than those using the another methods. Especially, the 160 basis vectors(features) of $16\times16$-pixel show the local features which have the characteristics of spatial frequency and oriented edges in the images.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1997.10a
• /
• pp.33-36
• /
• 1997

로봇이 지니는 지역적 한계성을 극복하기 위하여 구륜 이동용 로봇에 대한 연구가 전세계적으로 진행되어지고 있으나, 구륜이동로보트는 모델링의 불활실성이나 nonholomic등의 제약조건에 의하여 제어기의 설계시 많은 문제들을 지니게 된다. [1][2]. 이러한 어려움을 해결하기 위해 퍼지 알고리즘을 이용한 제어기 설계가 이루어지고 있으나 제한된 범위게 머무르고 있는 상황이다. 본 연구에서는 유전알고리즘에 근거하여 소속함수 및 규칙부의 자율적 조졸을 수행하는 구륜이동로보트의 퍼지 제어기를 한다. 제시된 알고리즘에서 퍼지 입출력 소속함수의 조절을 각각 독립적으로 이루어지며, 출력 소속함수의 유사지표에 근거하여 규칙부의 조절이 이루어진다.

• Annual Conference on Human and Language Technology
• /
• 2002.10e
• /
• pp.33-39
• /
• 2002

중국어의 품사 태깅(part-of-speech tagging)을 위해서는 중국어 문장들은 내부 단어간의 명확한 분리가 없기 때문에 단어 분할(word segmentation)과 품사 태깅을 동시에 처리해야 한다. 본 논문은 규칙 기반(rule base)과 사전 기반(dictionary base) 기법을 혼합하여 구현한 단어 분할 시스템을 사용하여 입력 문장을 단어 단위로 분할하고, HMM(hidden Markov model) 기반 통계적 품사 태깅 기법을 사용한다. 특히, 본 논문에서는 주어진 말뭉치(corpus)로부터 자동 학습(automatic training)을 통해 품사 사전을 구축하여 구현된 시스템과 말뭉치간의 독립성을 유지한다. 말뭉치는 중국어 간체와 번체 모두를 대상으로 하고, 각 말뭉치로부터 자동 학습을 통해 얻어진 품사 사전으로 단어 분할과 품사 태깅을 한다. 실험결과들은 간체, 번체 각각의 단어 분할 성능과 품사 태깅 성능을 보여준다.

• Korean Journal of Logic
• /
• v.18 no.2
• /
• pp.167-195
• /
• 2015

This paper aims to respond to Weisberg's claim that the standard Bayesian epistemology cannot model an agent's belief updating that is triggered by some undermining evidence. Our epistemological intuition seems to require that the undermining evidence decreases some particular relevant credences. According to Weisberg, however, such a belief change cannot result from the standard Bayesian belief updating rules-i.e., (Jeffrey) Conditionalization. This is because probabilistic independence between some propositions is preserved under (Jeffrey) Conditionalization on the relevant evidence. Yet I will show in this paper that this conclusion is somewhat hasty. In particular, I will show that there is another version of Conditionalization and that when one updates her credences by means of such a version, the belief updating originated in undermining evidence can be well modeled in the Bayesian framework. Some authors often call the version Higher Order Conditionalization.