• Annual Conference on Human and Language Technology
• /
• 2020.10a
• /
• pp.243-248
• /
• 2020

인공 신경망을 통한 심층 학습 모델의 발전은 컴퓨터 비전, 자연언어 이해 문제들에서 인간을 뛰어넘는 성능을 보이고 있다. 특히 트랜스포머[1] 기반의 사전 학습 모델은 질의응답, 대화문과 같은 자연언어 이해 문제에서 최근 높은 성능을 보이고 있다. 하지만 트랜스포머 기반의 모델과 같은 심층 학습 모델의 급격한 발전 양상에 비해, 이의 동작 방식은 상대적으로 잘 알려져 있지 않다. 인공 신경망을 통한 심층 학습 모델을 해석하는 방법으로 모델의 예측 값과 실제 값이 얼마나 일치하는지를 측정하는 모델의 보정(Calibration)이 있다. 본 연구는 한국어 기반의 심층학습 모델의 해석을 위해 모델의 보정을 수행하였다. 그리고 사전 학습된 한국어 언어 모델이 문장이 내포하는 애매성을 잘 파악하는지의 여부를 확인하고, 완화 기법들을 적용하여 문장의 애매성을 확신 수준을 통해 정량적으로 출력할 수 있도록 하였다. 또한 한국어의 문법적 특징으로 인한 문장의 의미 변화를 모델 보정 관점에서 평가하여 한국어의 문법적 특징을 심층학습 언어 모델이 잘 이해하고 있는지를 정량적으로 확인하였다.

• Korean Journal of Cognitive Science
• /
• v.28 no.4
• /
• pp.193-221
• /
• 2017

Grouping episodes into semantically related categories is necessary for better mnemonic structure. However, the effect of grouping on memory of subordinate details was not clearly understood. In an fMRI study, we tested whether attending superordinate during semantic association disrupts or enhances subordinate episodic details. In each cycle of the experiment, five cue words were presented sequentially with two related detail words placed underneath for each cue. Participants were asked whether they could imagine a category that includes the previously shown cue words in each cycle, and their confidence on retrieval was rated. Participants were asked to perform cued recall tests on presented detail words after the session. Behavioral data showed that reaction times for categorization tasks decreased and confidence levels increased in the third trial of each cycle, thus this trial was considered to be an important insight where a semantic category was believed to be successfully established. Critically, the accuracy of recalling detail words presented immediately prior to third trials was lower than those of followed trials, indicating that subordinate details were disrupted during categorization. General linear model analysis of the trial immediately prior to the completion of categorization, specifically the second trial, revealed significant activation in the temporal gyrus and inferior frontal gyrus, areas of semantic memory networks. Representative Similarity Analysis revealed that the activation patterns of the third trials were more consistent than those of the second trials in the temporal gyrus, inferior frontal gyrus, and hippocampus. Our research demonstrates that semantic grouping can cause memories of subordinate details to fade, suggesting that semantic retrieval during categorization affects the quality of related episodic memory.

• Journal of The Korean Association For Science Education
• /
• v.43 no.3
• /
• pp.307-319
• /
• 2023

This study aims to explain the key concepts and principles of text-based generative artificial intelligence (AI) that has been receiving increasing interest and utilization, focusing on its application in science education. It also highlights the potential and limitations of utilizing generative AI in science education, providing insights for its implementation and research aspects. Recent advancements in generative AI, predominantly based on transformer models consisting of encoders and decoders, have shown remarkable progress through optimization of reinforcement learning and reward models using human feedback, as well as understanding context. Particularly, it can perform various functions such as writing, summarizing, keyword extraction, evaluation, and feedback based on the ability to understand various user questions and intents. It also offers practical utility in diagnosing learners and structuring educational content based on provided examples by educators. However, it is necessary to examine the concerns regarding the limitations of generative AI, including the potential for conveying inaccurate facts or knowledge, bias resulting from overconfidence, and uncertainties regarding its impact on user attitudes or emotions. Moreover, the responses provided by generative AI are probabilistic based on response data from many individuals, which raises concerns about limiting insightful and innovative thinking that may offer different perspectives or ideas. In light of these considerations, this study provides practical suggestions for the positive utilization of AI in science education.

• Environmental and Resource Economics Review
• /
• v.14 no.1
• /
• pp.75-100
• /
• 2005

This study attempts to empirically investigate the respondents' preference uncertainty involved in stating their willingness to pay (WTP). In the contingent valuation (CV) survey, we employed two approaches using two split samples. The respondents of one sample were given the opportunity to express intensity of preference through polychotomous choice (PC) WTP question. Those of the other sample were given a follow-up question of confidence measure (0~100%). By incorporating the two elicited degrees of preference uncertainty into examining the WTP responses, we take a comparison of the two approaches in terms of the goodness-of-fit of the examination and the efficiency of the mean WTP estimates. In comparing the DC model with the PC models, the DC model provides more efficient estimates. Moreover, the conventional DC model give some gains in terms of the goodness-of-fit and efficiency in comparing with the PC model most similar to this model. In this specific study, incorporating the preference uncertainty in DC model results greater estimates than conventional DC model without loss of goodness-of-fit and efficiency. This implies that the consideration of preference uncertainty on DC model could correct underestimating. We conclude that DC model provides a better estimate of WTP and preference uncertainty could be a critical information on the DC-CV estimation.

• The Journal of the Korea Contents Association
• /
• v.14 no.5
• /
• pp.411-424
• /
• 2014

This study tries to establish the evaluation standards of the service quality of local cultural festivals in an effort to identify public awareness of those festivals. Also, based on the importance-performance importance-performance analysis on the service quality, this research is aimed at finding ways to enhance service quality and to improve operation plans in more efficient ways for the maximization of visitors' use of festivals. Ultimately, the current study seeks to make a contribution to improving the quality of local cultural festivals. The results of the correspondence t-test analysis of importance and performance revealed that significant differences were found in following items of five factors; 1) in the responsiveness factor, the presence or absence of rest space, facility guide, communication, and installation of public restroom/cleanliness 2) in the certainty factor, food prices and staff members' kindness 3) in the reliability factor, all items including festival pamphlets (to the festival venue), guidance facilities/signposts, hospitality, schedule information, and quick responses 4) in the tangibility factor as well, all items such as the awareness of local culture, benefits, and storytelling 5) in the empathy factor, diversity of events and festival contents.

• Proceedings of the Korea Technology Innovation Society Conference
• /
• 1998.05a
• /
• pp.14-14
• /
• 1998

기술추세분석(trend analysis)이나 특허분석(patent analysis) 등과 같이 객관적 자료를 이용하는 여타 기술예측방법과 달리, 델파이 방법은 해당 분야에 대한 전문가들의 식견(또는 주관적 평가)을 예측의 유일한 원천으로 하고 있다는 점에서 예측결과에 대한 신뢰도 분석을 강하게 요구하고 있다. 이러한 점에 착안하여 본 연구에서는 최근에 실시된 산업기술예측(1998) 자료를 이용하여 델파이 방법을 이용한 기술예측의 신뢰도 분석을 실시하였다. 이러한 분석은 전문가들의 예측결과에 대해 내적 일관성의 유지 여부와, 전문도가 상이한 집단간에 예측 결과의 차이가 있는지 여부를 중심으로 이루어졌으며 그 결과는 다음과 같다. 첫째, 응답 결과에 내적 일관성이 있는지 여부를 검정한 결과 생물·정밀화학분야를 제외한 모든 분야에서 일관성을 지니고 있는 것으로 나타나고 있다. 두번째, 전문도가 높은 응답자들일수록 자신의 예측결과를 확신하고는 있으나, 전문도가 낮은 응답자들에 비해 예측결과에 이견이 큰 것으로 분석되었으며, 마지막으로 전문도가 높은 응답자들과 그렇지 않은 응답자들 사이에 기술과제의 예상 실현시기에 대해서는 거의 차이가 없는 반면, 합의(consensus)의 정도는 큰 차이가 있는 것으로 나타났다. 이러한 분석결과는 델파이 방법을 이용한 기술예측을 설계하는 데 있어 몇가지 시사점을 제공하고 있다. 첫번째로 기술예측시 전문가들의 예측결과에 대해 내적 일관성이 존재하는 지 여부를 검증할 수 있도록 예측과정을 설계하는 것이 바람직하다는 것이다. 이러한 설계과정은 델파이 방법이 예측결과를 검증할 만한 객관적인 장치를 지니지 못하고 있다는 점에서 더욱 필요하다고 하겠다. 두번째는 정보로서의 가치가 큰 전문도가 높은 응답자들의 예측결과를 활용하기 위해서는, 예측결과에 대한 이들의 합의(consensus)의 정도를 높일 수 있는 방안이 마련되어야 한다는 것이다. 델파이의 최종 라운드가 진행된 이후 이들에 대해서만 추가적인 라운드를 실시하거나, 예측과정에서 이들에게 관련 정보를 제공하는 것도 하나의 대안이 될 수 있을 것이다.

• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.9
• /
• pp.675-684
• /
• 2009

Internet worm can cause a traffic problem through DDoS(Distributed Denial of Services) or other kind of attacks. In those manners, it can compromise the internet infrastructure. In addition to this, it can intrude to important server and expose personal information to attacker. However, current detection and response mechanisms to worm have many vulnerabilities, because they only use local characteristic of worm or can treat known worms. In this paper, we propose a new framework to detect unknown worms. It uses macroscopic characteristic of worm to detect unknown worm early. In proposed idea, we define the macroscopic behavior of worm, propose a worm detection method to detect worm flow directly in IP packet networks, and show the performance of our system with simulations. In IP based method, we implement the proposed system and measure the time overhead to execute our system. The measurement shows our system is not too heavy to normal host users.

• Korean Journal of Construction Engineering and Management
• /
• v.8 no.4
• /
• pp.137-145
• /
• 2007

Purpose of this study is to analyze degrees of oral transmission, customer satisfaction, and subscription customer's re-visitation to grasp effectiveness of CRM of construction company. Brand recognition, customer satisfaction, and customer loyalty can be made to improve by CRM of construction company. Ultimately, this study is intended to improve management achievement by increasing total sales and reducing expense. 175 distribution customers and subscription customers participate in the final analysis for this study. In terms of service quality factors, state of facility, trust, response, conviction and sympathy are measured. In terms of effectiveness, customer satisfaction, oral transmission, and customer's re-visitation are measured. Considering these things, CRM of construction company can improve oral transmission, satisfaction, re-visitation of distribution customers and subscription customers. CRM of construction company can be utilized as a competitive means for management and contribute to improving management achievement in the long run. Through examining actual effectiveness of CRM of construction company in this study, it can be thought that necessity and possibility of CRM of construction company should be recognized and CRM of construction company should be developed further.

• The Journal of the Convergence on Culture Technology
• /
• v.7 no.3
• /
• pp.231-239
• /
• 2021

The purpose of this study is to verify the effectiveness of service quality on satisfaction factors, and to provide practical information on the management of Taekwondo Dojang nationwide, which are facing difficulties due to the decrease in trainees and COVID-19. In this study, 11 studies were selected from January 2010 to December 2020. The conclusion is as follows. First of all, it has been confirmed that the overall influence of Taekwondo dojo service on satisfaction is moderate. Second, the effectiveness of low-quality service on satisfaction (painting, customers, trainees) was the highest, and there was no significant difference between the factors in order of confidence, response, empathy and reliability. Third, the quality, effectiveness and satisfaction of the satisfaction type were the greatest, and the satisfaction of the trainees was the coating and customer satisfaction.

• 한국노년학
• /
• v.31 no.3
• /
• pp.691-709
• /
• 2011

The purpose of this study is to measure the utilization of a day care service and service quality by SERVQUAL model, also to discuss policy implications based on this. The data analysis is conducted with 208 users of a day care service in Deajeon metropolitan city, Chungnam and Chungbuk area. The analysis result is as follows. First, it shows positive definiteness on tangibles, reliability, responsiveness, assurance, empathy in all categories because perceived level is higher than expectations. Perceived level after using the service is more highly rated than expectations before using it by most of seniors who use the service. Second, service quality is examined in accordance with demographic characteristics, but the difference is not statistically significant. Third, examining service quality in terms of characteristics of centers, it shows statistically significant difference in perceived level depending on management type of location, a number of provided programs per week, and the size of a facility and in expectations depending on established institution. Forth, as a result of examining service quality in accordance with characteristics of service users, there is no meaningful difference statistically.