• Annual Conference on Human and Language Technology
• /
• 한국정보과학회언어공학연구회 2020년도 제32회 한글 및 한국어 정보처리 학술대회
• /
• pp.501-506
• /
• 2020

질의 난이도 분석 문제는 자연어 질의문을 답변할 때 어려움의 정도를 측정하는 문제이다. 질의 난이도 분석 문제는 문서 독해, 의학 시험, 비디오 질의 등과 같은 다양한 데이터셋에서 연구되어 왔다. 본 논문에서는 질의문과 질의문에 응답하기 위한 정보들 간의 관계를 파악하는 것으로 질의 난이도 분석 문제를 접근하여 이를 BERT와 Dual Multi-head Attention을 사용하여 모델링 하였다. 본 논문에서 제안하는 모델의 우수성을 증명하기 위하여 최근 자연언어이해 부분에서 높은 성능을 보여주는 기 학습 언어 모델과 이전 연구의 질의 난이도 분석 모델과의 성능을 비교하였고, 제안 모델은 대표적인 비디오 질의 응답 데이터셋인 DramaQA의 Memory Complexity에서 99.76%, Logical Complexity에서는 89.47%의 정확도로 가장 높은 질의 난이도 분석 성능을 보여주었다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 한국컴퓨터정보학회 2019년도 제59차 동계학술대회논문집 27권1호
• /
• pp.127-128
• /
• 2019

컴퓨터 프로그래밍 수업에 참여하는 학습자들에게 성찰일지를 활용하여 수업에서 배운 내용을 즉시 복습하고 교수자와의 1:1 질의 응답 활동을 진행한 결과, 성찰일지를 통해 가장 도움이 받은 학습 영역과 얻은 학습 성과가 무엇인지 학습자 설문을 통해 조사하고 분석해보았다. 이 연구를 통해 프로그래밍 학습자들을 위해 효과적인 학습 활동을 지원함으로써 난이도가 높은 프로그래밍 수업에서 발생하는 학습자들의 학습 능력 차이를 줄이고 학습 만족도를 높일 수 있을 것으로 기대된다.

• The Journal of Korean Academy of Sensory Integration
• /
• 제21권2호
• /
• pp.45-57
• /
• 2023

Objective : The purpose of this study was to investigate the concurrent validity of the self- and proxy-report versions of the KIDSCREEN-10 quality of life questionnaire. Methods : A total of nine children and nine parents were selected to represent a cohort registered for a school-based wellness program. Two versions of the KIDSCREEN-10 questionnaire (self- and proxy reports) were administered to the children and their parents. The Rasch rating scale model was applied to determine the dimensionality and item difficulty of the two versions of the questionnaire. Moreover, the item-person matching map and Spearman's rho were compared to confirm the concurrent validity of the two versions. Results : All items, except four items (i.e., autonomy, home life, concentration/learning, and peers/social support), fit the Rasch rating scale model of the children's self-report version of the questionnaire. With regard to the parent's proxy-report version, two items misfit the model. While the items of the self- and proxy-report versions showed similar item difficulties, the parents had a tendency to be more severe in their ratings than the children. The correlation between the two versions was relatively low (Spearman's rho = .533, p > .05). The scatterplots between the two versions showed differences in the item difficulties of the physical and psychological well-being and self-perception items. Conclusion : These findings suggest that the three identified items should be taken into consideration when measuring children's health-related quality of life using the KIDSCREEN-10 questionnaire.

• The Journal of Korean Association of Computer Education
• /
• 제9권6호
• /
• pp.11-18
• /
• 2006

The purpose of this paper is to draw problems from analyzing "General Computer" questions of Career Searching Section in the College Scholastic Ability Test in 2005 and 2006 and to offer some suggestions about them. For the qualitative research, this paper analyzed content validity. For the quantitative research, this paper analyzed item difficulty and item discrimination by using Bayesian 1.0 based on 2-parameter item response model and this paper analyzed item reliability and distracters by using Testan 1.0. By analyzing "General Computer" questions, this paper would like to improve the quality of items and estimate item difficulty. Therefore, "General Computer" questions could be suggested as materials for developing reliable and discriminative questions.

• The korean journal of orthodontics
• /
• 제35권1호
• /
• pp.60-68
• /
• 2005

The evaluation of malocclusion has to be done quantitatively and qualitatively. This will be lead toward an analysis of malocclusion severity as well as treatment difficulty. The method of proper evaluation of malocclusion severity and treatment difficulty is necessary to assess treatment effect and efficiency for the orthodontists and to establish fundamentals for planning and executing the health-related policies in private and public institutions. The purposes of this study as the first part of the objective and quantitative analysis of malocclusion were 1) to measure treatment difficulty based on the opinions of several orthodontists. and 2) to investigate the relationships between objective malocclusion severity and subjective treatment difficulty 100 pairs of dental casts that had various types and severity of malocclusion were selected from the orthodontic departments of Kyurghee University and Samsung Medical Center The objective malocclusion severity was measured with the PAR (Peer Assessment Rating) index and the subjective treatment difficulty was evaluated by 8 experienced orthodontists. The relationships between objective malocclusion severity and subjective treatment difficulty were statistically evaluated. There were significant relationships between objective malocclusion severity and subjective treatment difficulty especially in the measurements of the upper anterior alignment, the buccal occlusion. the overjet, the overbite and the midline discrepancy en the malocclusion components. The results of this study can provide the background knowledge to develop a new occlusal index. which contains both the malocclusion severity and treatment difficulty for Korean orthodontists.

• Journal of Digital Contents Society
• /
• 제8권4호
• /
• pp.531-539
• /
• 2007

An advance in digital technologies and an introduction of HDTV contents production systems have made innovative changes in a TV broadcasting program and video production environment. The role of HD contents production experts is particularly getting more important to have a competitiveness in digital video production. It is necessary to develop a professional training and reeducation program that is based upon the perception of production experts. In this paper, we ed the perception of video production professionals and analyzed their requirements. With a DACUM job analysis method, 9 job duties and 71 tasks were retrieved to establish a research model. Then the performing frequency, importance, and difficulty in video editor's duties were analyzed and discussed.

• Journal of the Korean earth science society
• /
• 제26권6호
• /
• pp.511-523
• /
• 2005

The purpose of this study is to analyze the 1st and 2nd Korea Earth Science Olympiad (KESO) items, in order to find informations to enhance item quality. To do this, internal and external item classification frameworks are developed. Item difficulty (P), discrimination index (DI), correlation, and reliability are estimated by using classical test theory. Generalizability is also estimated by applying the generalizability theory. The results of item classification are as follows: (1) ‘Geology’, ‘astronomy’ and ‘data analysis and interpretation’ are dominant in content and inquiry process domain, respectively. Nearly every item has textbook context. (2) There is no difference between the preliminary and final tests in terms of their thinking skills sections. (3) As a whole, the ratio of items with pictures is high in item representation. However, multiple-choice and short answer items are more common in preliminary competition, and essay type items are found more often in final competition. The ratio of simple items is high in middle school section and preliminary competition, but composite items are dominant in high school section and final competition. The findings of item analysis are as follows: (1) In the middle school section, P is low and DI is moderate. But in the high school section, there is a considerable differences between science high schools and other high schools in general. (2) The highest correlation is reported between the scores of meteorology domain and total score in middle school, whereas in high school astronomy domain and total score show the highest correlation. (3) General high school section show the highest Cronbach $\alpha$ and generalizability. (4) General high school section show acceptable generalizability coefficient (> 0.80), but middle and science high school section should increase the number of items to reach acceptable generalizability level.

• Journal of Practical Engineering Education
• /
• 제6권1호
• /
• pp.43-49
• /
• 2014

It is known that the ability and habit to "flow" could be one of the important indices to evaluate one's life. And the flow experience of university students who are in the important time transferring from the adolescence to the initial adult period will give an important effect firstly to study one's major and map out one's career path, finally to spread out the high quality life. This study investigated the university students who are in School of Mechatronic Engineering to experience the flow in the learning of PLC (Programmable Logic Controller) practice. For this, the flow condition was arranged. And in order to experience the flow in the learning of PLC, lecture plan was established, studying environment was changed and level of difficulty was controlled. Then the implications of the flow were explored by a 5-point Likert scale survey performed to 90 students.

• Annual Conference on Human and Language Technology
• /
• 한국정보과학회언어공학연구회 2004년도 제16회 한글.언어.인지 한술대회
• /
• pp.275-282
• /
• 2004

본 논문은 3단계 정답 추출 방법을 통해 백과사전 인물분야 질의응답 시스템을 구현하는 방법을 제안한다. 논문에서 제안한 3단계 정답 추출 방법은 1) 백과사전 문서 내에서 정형화 될 수 있는 지식들을 추출한 백과사전 KB 기반 정답 추출 방법, 2) 문장을 언어분석 하여 LF(Logical Form)구조를 추출하여 색인한 LF 기반 정답추출 방법, 3) 각 문장을 주제 태깅을 하여, 주제별로 묶어 의미적 단락으로 구분하고 단락 검색을 기반으로 정답을 추정하는 의미적 단락 기반 정답 추출 방법으로 구성되어 있다. 이러한 방법론은 백과사전이라는 문서 도메인의 특성을 반영하고. 사용자 질문의 난이도 또는 형태에 따라서 정답을 제공할 수 있는 백과사전 인물분야 질의응답 시스템에 적합하다.

• Journal of the Korea Computer Industry Society
• /
• 제3권8호
• /
• pp.1007-1018
• /
• 2002

A systematic and multi-dimensional analysis in evaluation results may play a role of providing both learners and instructors with essential information. Conventional types of evaluation research have a tendency of partiality for developing evaluation tools, and analyzing the evaluation results has mostly been fragmentary and a one-dimensional analysis. In this study, through analyzing evaluation results in a multi-dimensional way, a multi-dimensional evaluation result analyzing system was developed for the purpose of providing various information for both learners and instructors to accomplish quality learning teaching. Dimensions are classified into four dimensions including period, student, difficulty degree, and evaluation domain. Analysis results are presented in various types of Dcube, graphs, and spreadsheets.