• Journal of People, Plants, and Environment
• /
• v.22 no.4
• /
• pp.341-354
• /
• 2019

Urban agriculture, which promotes communication in vulnerable classes and the formation of social networks has been gaining attention with an emphasis on healthy city, elderly-friendly city, safe city and happy city as future keywords about urban life. There is a growing interest in public awareness in many areas such as health, society, economy, and ecology. As an attempt to improve the diversity of urban gardens, this study begins with collecting suitable crops for urban gardens from "Imwongyeongjeji (林園經濟志)," an encyclopedia written by Yoo-Ku Seo, a scholar in the 18-19th century. Out of those recorded in "Gwanhyuji (灌畦志)," 128 kinds of crops with linkage of the historical achievements of the realists who gave their priority to public welfare were selected and 53 crops which had traditionality, historicality, health functionality and popularity were finally selected. The properties (cold, warm, clam) of the selected crops were evenly distributed, and there was no crop that was hot and cool. In addition, the number of crops that have a sweet taste was the highest, followed by spicy and bitter, but there was no salty vegetable, which can be attributed to the fact that 12 namuls (wild vegetables) that grow in seas were excluded in this study since they were not suitable for urban gardens. Urban gardens can be transformed from those that focus on primary production and secondary consumption activities into a new resource that offers educational and traditional values by applying humanities to urban agriculture as a content resource in the era of cultural consilience and convergence. It is expected to satisfy urban residents' intellectual and participatory needs and to enhance the diversity and utility of urban gardens by applying traditional knowledge to a new model of urban agriculture. We hope that further research will be conducted to develop new types and models of urban agriculture going forward.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.483-483
• /
• 2022

댐과 저수지의 운영 최적화를 위한 수문학적 예보는 현재 수동적인 댐 운영이 주를 이루면서 활용도가 높지 않다. 불확실한 기후변화나 기후재난 상황에서 우리 사회에 악영향을 최소화하기 위해 선제적으로 대응/대비할 수 있는 댐 운영 방안이 불가피하다. 강우량 예측 기술은 기후변화로 인해 제한적인 상황이다. 실례로, 2020년 8월에 섬진강의 댐이 극심한 집중 강우로 인해 무너지는 사태가 발생하였고 이로 인해 지역사회에 막대한 경제적 피해가 발생하였다. 선제적 댐 방류량 운영 기술은 또한 환경적인 변화로 인한 영향을 완화하기 위해 필요한 것이다. 제한적인 기상 예보 기술을 극복하고자 심화학습이나 강화학습 같은 인공지능 모델들의 활용성에 대한 연구가 시도되고 있다. 따라서 본 연구는 섬진강 댐의 시간당 수문 데이터를 이용하여 댐 운영을 위한 심화학습 모델을 개발하고 그 활용도를 평가하였다. 댐 운영을 위한 심화학습 모델로서 시계열 데이터 예측에 적합한 Long Sort Term Memory(LSTM)과 Gated Recurrent Unit(GRU) 알고리즘을 구축하고 댐 수위를 예측하였다. 분석 자료는 WAMIS에서 제공하는 2000년부터 2021년까지의 시간당 데이터를 사용하였다. 입력 데이터로서 시간당 유입량, 강우량과 방류량을, 출력 데이터로서 시간당 수위 자료를 각각 사용하였으며. 결정계수(R2 Score)를 통해 모델의 예측 성능을 평가하였다. 댐 수위 예측값 개선을 위해 하이퍼파라미터의 '최적값'이 존재하는 범위를 줄여나가는 하이퍼파라미터 최적화를 두 가지 방법으로 진행하였다. 첫 번째 방법은 수동적 탐색(Manual Search) 방법으로 Sequence Length를 24, 48, 72시간, Hidden Layer를 1, 3, 5개로 설정하여 하이퍼파라미터의 조합에 따른 LSTM와 GRU의 민감도를 평가하였다. 두 번째 방법은 Grid Search로 최적의 하이퍼파라미터를 찾았다. 이 두가지 방법에서는 같은 하이퍼파라미터 안에서 GRU가 LSTM에 비해 더 높은 예측 정확도를 보였고 Sequence Length가 높을수록 정확도가 높아지는 경향을 보였다. Manual Search 방법의 경우 R2가 최대 0.72의 정확도를 보였고 Grid Search 방법의 경우 R2가 0.79의 정확도를 보였다. 본 연구 결과는 가뭄과 홍수와 같은 물 재해에 사전 대응하고 기후변화에 적응할 수 있는 댐 운영 개선에 도움을 줄 수 있을 것으로 판단된다.

• The Journal of Health Technology Assessment
• /
• v.1 no.1
• /
• pp.22-26
• /
• 2013

Objectives: Balancing benefits and risks through the drug life cycle has been discussed for many decades. The objective of this study was to review the processes and tools currently proposed for benefit-risk assessment of medicinal drugs. It aimed to establish scientific and efficient drug safety management system based on the synthetic analysis of benefit-risk evidence. Methods: We conducted a review of exiting literatures published by regulatory agencies or initiatives. Not only quantitative methodologies but also qualitative method were compared to understand their key characteristics for the benefit and risk assessment of drugs. Results: Recently, benefit-risk assessments have more structured approaches to decision making as part of regulatory science. Regulatory agencies such as European Medicines Agency, FDA have prepared plans to apply benefit-risk assessment to regulatory decision making. Also many initiatives such as IMI (Innovative Medicine Initiative) have conducted research and published reports about benefit-risk assessment. For benefit-risk assessment, four kinds of methods are necessary. Frameworks such as BRAT (Benefit Risk Action Team) framework, PrOACT-URL provide guidance for the whole process of decision-making. Metrics are measurements of risk benefit. The estimation techniques are methods to synthesis and combine evidences from various sources. The utility survey techniques are necessary to explicit preferences of various outcome from stakeholders. Conclusion: There is the lack of widely accepted, validated model for benefit-risk assessment. Nor there is an agreement among academia, industry, and government on methods for the quantitative valuation. It is also limited by available evidence and underlying assumptions. Nevertheless, benefit-risk assessment is fundamental to improve transparency, consistency and predictability for decision making through the structured systematic approaches.

• Journal of KIBIM
• /
• v.13 no.3
• /
• pp.21-29
• /
• 2023

ChatGPT, a chatbot based on GPT large language models, has gained immense popularity among the general public as well as domain professionals. To assess its proficiency in specialized fields, ChatGPT was tested on mainstream exams like the bar exam and medical licensing tests. This study evaluated ChatGPT's ability to answer questions related to Building Information Modeling (BIM) by testing it on Korea's BIM expertise exam, focusing primarily on multiple-choice problems. Both GPT-3.5 and GPT-4 were tested by prompting them to provide the correct answers to three years' worth of exams, totaling 150 questions. The results showed that both versions passed the test with average scores of 68 and 85, respectively. GPT-4 performed particularly well in categories related to 'BIM software' and 'Smart Construction technology'. However, it did not fare well in 'BIM applications'. Both versions were more proficient with short-answer choices than with sentence-length answers. Additionally, GPT-4 struggled with questions related to BIM policies and regulations specific to the Korean industry. Such limitations might be addressed by using tools like LangChain, which allow for feeding domain-specific documents to customize ChatGPT's responses. These advancements are anticipated to enhance ChatGPT's utility as a virtual assistant for BIM education and modeling automation.

• Journal of the Korean Geotechnical Society
• /
• v.40 no.2
• /
• pp.81-93
• /
• 2024

In Korea, numerous large-scale infrastructure construction projects and housing site developments are being undertaken. However, due to limited land availability, sourcing high-quality backfill materials that meet the standards for railroad and road embankment compaction and mechanically stabilized earth (MSE) retaining wall construction poses significant challenges. Concurrently, there has been an increase in structural failures of many MSE retaining walls, attributed primarily to reduced bearing capacity and impaired drainage performance, resulting from inadequate backfill compaction. This study aimed to analyze the structural performance and safety of an MSE retaining wall using recycled soil as backfill. We conducted small-scale model tests utilizing 3D printing technology combined with two-dimensional numerical analysis. The study quantitatively evaluated the MSE retaining wall's performance concerning the recycled soil mixing ratio and reinforcement installation methods. Furthermore, the utility of 3D printing was confirmed through the production of an experimental wall designed to facilitate easy reinforcement attachment, mirroring the conditions of actual MSE retaining wall construction.

• Journal of Korean Neurosurgical Society
• /
• v.67 no.2
• /
• pp.166-176
• /
• 2024

Objective : Three-dimensional (3D) printing in vascular surgery is trending and is useful for the visualisation of intracranial aneurysms for both surgeons and trainees. The 3D models give the surgeon time to practice before hand and plan the surgery accordingly. The aim of this study was to examine the effect of preoperative planning with 3D printing models of aneurysms in terms of surgical time and patient outcomes. Methods : Forty patients were prospectively enrolled in this study and divided into two groups : groups I and II. In group I, only the angiograms were studied before surgery. Solid 3D modelling was performed only for group II before the operation and was studied accordingly. All surgeries were performed by the same senior vascular neurosurgeon. Demographic data, surgical data, both preoperative and postoperative modified Rankin scale (mRS) scores, and Glasgow outcome scores (GOS) were evaluated. Results : The average time of surgery was shorter in group II, and the difference was statistically significant between the two groups (p<0.001). However, no major differences were found for the GOS, hospitalisation time, or mRS. Conclusion : This study is the first prospective study of the utility of 3D aneurysm models. We show that 3D models are useful in surgery preparation. In the near future, these models will be used widely to educate trainees and pre-plan surgical options for senior surgeons.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.47 no.2
• /
• pp.134-146
• /
• 2024

This study introduces and experimentally validates a novel approach that combines Instruction fine-tuning and Low-Rank Adaptation (LoRA) fine-tuning to optimize the performance of Large Language Models (LLMs). These models have become revolutionary tools in natural language processing, showing remarkable performance across diverse application areas. However, optimizing their performance for specific domains necessitates fine-tuning of the base models (FMs), which is often limited by challenges such as data complexity and resource costs. The proposed approach aims to overcome these limitations by enhancing the performance of LLMs, particularly in the analysis precision and efficiency of national Research and Development (R&D) data. The study provides theoretical foundations and technical implementations of Instruction fine-tuning and LoRA fine-tuning. Through rigorous experimental validation, it is demonstrated that the proposed method significantly improves the precision and efficiency of data analysis, outperforming traditional fine-tuning methods. This enhancement is not only beneficial for national R&D data but also suggests potential applicability in various other data-centric domains, such as medical data analysis, financial forecasting, and educational assessments. The findings highlight the method's broad utility and significant contribution to advancing data analysis techniques in specialized knowledge domains, offering new possibilities for leveraging LLMs in complex and resource-intensive tasks. This research underscores the transformative potential of combining Instruction fine-tuning with LoRA fine-tuning to achieve superior performance in diverse applications, paving the way for more efficient and effective utilization of LLMs in both academic and industrial settings.

• Journal of Global Scholars of Marketing Science
• /
• v.19 no.2
• /
• pp.17-26
• /
• 2009

As the lifestyle of consumers changes and the need for various products increases, new products are being developed in the market. Each household owns various home appliances which are purchased through the choice of a decision maker. These appliances include not only large-sized products such as TV, refrigerator, and washing machine, but also small-sized products such as microwave oven and air cleaner. There exists latent correlation among possession of home appliances, even though they are purchased independently. The purpose of this research is to analyze the effect of demographic factors on the purchase and possession of each home appliances, and to derive some relationships among various appliances. To achieve this purpose, the present status on the possession of each home appliances are investigated through consumer survey data on the electric and energy product. And a multivariate probit(MVP) model is applied for the empirical analysis. From the estimation results, some appliances show a substitutive or complementary pattern as expected, while others which look apparently unrelated have correlation by co-incidence. This research has several advantages compared to previous literatures on home appliances. First, this research focuses on the various products which are purchased by each household, while previous researches such as Matsukawa and Ito(1998) and Yoon(2007) focus just on a particular product. Second, the methodology of this research can consider a choice process of each product and correlation among products simultaneously. Lastly, this research can analyze not only a substitutive or complementary relationship in the same category, but also the correlation among products in the different categories. As the data on the possession of home appliances in each household has a characteristic of multiple choice, not a single choice, a MVP model are used for the empirical analysis. A MVP model is derived from a random utility model, and has an advantage compared to a multinomial logit model in that correlation among error terms can be derive(Manchanda et al., 1999; Edwards and Allenby, 2003). It is assumed that the error term has a normal distribution with zero mean and variance-covariance matrix ${\Omega}$. Hence, the sign and value of correlation coefficients means the relationship between two alternatives(Manchanda et al., 1999). This research uses the data of 'TEMEP Household ICT/Energy Survey (THIES) 2008' which is conducted by Technology Management, Economics and Policy Program in Seoul National University. The empirical analysis of this research is accomplished in two steps. First, a MVP model with demographic variables is estimated to analyze the effect of the characteristics of household on the purchase of each home appliances. In this research, some variables such as education level, region, size of family, average income, type of house are considered. Second, a MVP model excluding demographic variables is estimated to analyze the correlation among each home appliances. According to the estimation results of variance-covariance matrix, each households tend to own some appliances such as washing machine-refrigerator-cleaner-microwave oven, and air conditioner-dish washer-washing machine and so on. On the other hand, several products such as analog braun tube TV-digital braun tube TV and desktop PC-portable PC show a substitutive pattern. Lastly, the correlation map of home appliances are derived using multi-dimensional scaling(MDS) method based on the result of variance-covariance matrix. This research can provide significant implications for the firm's marketing strategies such as bundling, pricing, display and so on. In addition, this research can provide significant information for the development of convergence products and related technologies. A convergence product can decrease its market uncertainty, if two products which consumers tend to purchase together are integrated into it. The results of this research are more meaningful because it is based on the possession status of each household through the survey data.

• The Korean Journal of Food And Nutrition
• /
• v.22 no.1
• /
• pp.141-149
• /
• 2009

The elderly population in Korea is growing rapidly and their needs for long-term care has also increased. By the year 2018, our society will be approaching aged society and by 2026 it will be a super-aged society. The purpose of this study was to employ conjoint analysis to establish the relative importance of foodservice encounters in terms of determining the utility values of hospital foodservice for elderly patients. According to the results pearson's R(0.420) and Kendall's tau(0.402) statistics showed that the model fits the data well(p<0.05). The relative importance scores of hospital foodservice encounters were as follows: dietary counseling with dietetics(51.2%), foodservice personnel(48.7%), and food(0.1%). A soft cooking method(0.001) was preferred to a general cooking method(0.001), and kind foodservice personnel(0.086) were preferred to quick service(-0.086). Finally, counseling with a dietitian once a week(-0.138) was preferred to counseling twice a week (-0.276). Based on this conjoint analysis, the most preferable model for foodservice at a long-term care facility would be; soft cooking methods, kind service by foodservice personnel, and dietetic counseling once a week. Overall, a better understanding of the specific needs of our institutionalized elderly is one of the key elements that can help our long-term care system develop improved foodservice programs.

• Journal of Global Scholars of Marketing Science
• /
• v.18 no.1
• /
• pp.141-156
• /
• 2008

Market structure analysis is a very useful tool to analyze the competition boundary of the brand or the company. But most of the studies in market structure analysis, the concern lies in nondurable goods such as candies, soft drink and etc. because of the their availability of the data. In the field of durable goods, the limitation of the data availability and the repurchase time period constrain the study. In the analysis of the automobile market, those of views might be more persuasive. The purpose of this study is to analyze the structure of automobile market based on some idea suggested by prior studies. Usually the buyers of the automobile tend to buy upper tier when they buy in the next time. That kind of behavior make it impossible to analyze the structure of automobile market under the level of automobile model. For that reason I tried to analyze the market structure in the brand or company level. In this study, consideration data was used for market structure analysis. The reasons why we used the consideration data are summarized as following. Firstly, as the repurchase time cycle is too long, brand switching data which is used for the market analysis of nondurable good is not avaliable. Secondly, as we mentioned, the buyers of the automobile tend to buy upper tier when they buy in the next time. We used survey data collected in the U.S.A. market in the year of 2005 through questionaire. The sample size was 8,291. The number of brand analyzed in this study was 9 among 37 which was being sold in U.S.A. market. Their market share was around 50%. The brands considered were BMW, Chevrolet, Chrysler, Dodge, Ford, Honda, Mercedes, and Toyota. �� ratio was derived from frequency of the consideration set. Actually the frequency is different from the brand switch concept. In this study to compute the �� ratio, the frequency of the consideration set was used like a frequency of brand switch for convenience. The study can be divided into 2 steps. The first step is to build hypothetical market structures. The second step is to choose the best structure based on the hypothetical market structures, Usually logit analysis is used for the choice best structure. In this study we built 3 hypothetical market structure. They are type-cost, cost-type, and unstructured. We classified the automobile into 5 types, sedan, SUV(Sport Utility Vehicle), Pickup, Mini Van, and Full-size Van. As for purchasing cost, we classified it 2 groups based on the median value. The median value was $28,800. To decide best structure among them, maximum likelihood test was used. Resulting from market structure analysis, we find that the automobile market of USA is hierarchically structured in the form of 'automobile type - purchasing cost'. That is, result showed that automobile buyers considered function or usage first and purchasing cost next. This study has some limitations in the analysis level and variable selection. First, in this study only type of the automobile and purchasing cost were as attributes considered for purchase. Considering other attributes is very needful. Because of the attributes considered, only 3 hypothetical structure could be analyzed. Second, due to the data, brand level analysis was tried. But model level analysis would be better because automobile buyers consider model not brand. To conduct model level study more cases should be obtained. That is for acquiring the better practical meaning, brand level analysis should be conducted when we consider the actual competition which occurred in the real market. Third, the variable selection for building nested logit model was very limited to some avaliable data. In spite of those limitations, the importance of this study lies in the trial of market structure analysis of durable good.