• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.410-413
• /
• 2021

본 논문에서는 분류 모델의 주류로 사용되고 있는 Encoder 기반 사전학습 모델(BERT, ALBERT, ELECTRA)의 내부 Encoder Layer가 하부 Layer에서는 Syntactic한 분석을 진행하고 상부 Layer로 갈수록 Semantic 한 분석을 진행하는 점, Layer가 구성됨에 따라 Semantic 정보가 Syntactic 정보를 개선해 나간다 점에 기반한 기존 연구 결과를 바탕으로 Encoder Layer를 구성함에 따라 어떻게 성능이 변화하는지 측정한다. 그리고 의도 분류를 위한 학습 데이터 셋도 분류하고자 하는 성격에 따라 Syntactic한 구성과 Semantic한 구성을 보인다는 점에 착안하여 ALBERT 및 ELECTRA를 이용한 의도 분류 모델을 구축하고 각 데이터 셋에 맞는 최적의 Encoder Layer 구성을 가지는 모델을 비교한 결과, 두 데이터 셋 간에 다른 Layer 구성을 보이는 점과 기존 모델보다 성능이 향상됨을 확인하였다.

• Journal of Embryo Transfer
• /
• v.29 no.2
• /
• pp.189-194
• /
• 2014

The reproductive disorders are the major causes of reproductive infertility in cows that affect the total annual calf crop, resulting in great economic loss in Bangladesh. The aim of the study was to find out the reproductive disorders (RD) in dairy cows that markedly influences the reproductive performances in aspect of Bangladesh. A total number of 1658 dairy cows were selected according to their body condition score (BCS) in different farms at the southern part of Bangladesh during the period of 2011 to 2012. The preliminary data (basic information) were collected directly from the dairy farmer's record books and asking questions according to a prescribed questionnaires as well as the diagnosis of RD was presumptively confirmed on the basis of history, clinical signs and examination of animals by ultrasonography and others necessary tools. There are thirteen major reproductive disorders were identified. Overall prevalence of reproductive disorders at that area were 23%, among of these anoestrus 5.1%, repeat breeder 3.7%, metritis 4.4%, poor heat detection 1.6%, ovarian cyst 0.36%, retain placenta 4.6%, dystocia 0.97% and pyometra 0.24%. It is indicated that anoestrus and retention of placenta after calving was most hazardous cause of infertility whereas the metritis and repeat breeder were the second line of consequence. RD had shown significantly higher incidence in low BCS (${\leq}2$) than that of fair (2.5) and very good (${\geq}3{\sim}3.5$). In conclusion, the highest RD especially anoestrus and retention of placenta is very alarming for reproductive loss which might be needed further research to identify the specific cause of these disorders for establishment a profitable dairying and dairy population.

• Journal of Korean Society of Transportation
• /
• v.32 no.4
• /
• pp.380-388
• /
• 2014

Annual average daily traffic(AADT) serves as important basic data in the transportation sector. AADT is used as design traffic which is the basic traffic volume in transportation planning. Despite of its importance, at most locations, AADT is estimated using short term traffic counts. An accurate AADT is calculated through permanent traffic counts at limited locations. This study dealt with estimating AADT using various models considering both the spatial correlation and time series data. Kriging models which are commonly used spatial statistics methods were applied and compared with each model. Additionally the External Universal kriging model, which includes explanatory variables, was used to assure accuracy of AADT estimation. For evaluation of various kriging methods, AADT estimation error, proposed using national highway permanent traffic count data, was analyzed and their performances were compared. The result shows the accuracy enhancement of the AADT estimation.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.41 no.1
• /
• pp.70-83
• /
• 2018

Korea's five leading industries-chemical, automobile, shipbuilding, metal and electronics-have been the growth engines of the Korean economy for the last 30 years. However, with their performance weakening in the past few years, Korea's annual growth rate has dropped to less than 2%. Just as the Japanese economy struggled through the lost decades, Korea's leading industries have been facing signals of arriving at a Strategic Inflection Point ("SIP") since the mid-2000s. The objective of this study is to present an appropriate value strategy for Korean firms in leading industries to gain sustainable competitive advantages. This paper examines the following issues : First, it diagnoses the signs and timing of SIP for the five leading industries. Second, this study examines 78 Korean and Japanese firms in order to understand the adaptation strategies of the highest and lowest performing firms from a value strategy point of view. Third, it empirically analyzes the correlation between the indicators of value strategy and the accounting performances of the Korean and Japanese firms. The data set used in the cluster analysis were collected from KISLINE database of NICE Information Service and annual reports. The findings in the case analyses of high and low performing companies show that high value-added specialty chemical materials and electronic component manufacturers that had a strong focus on value creation were defined as high performing companies. And in the cosmetic and automobile industries, companies with a strong emphasis on value appropriation gained high performance. Furthermore, the study conducted a cluster analysis based on two indicators R&D and advertisement expense. The analysis shows that a significant gap exists in the operating incomes among three groups classified under the value strategy. The study recommends that Korean firms are requested to increase R&D expenditure, conduct M&A and collaborate with companies that own source technologies to narrow the technology gap and improve their segment portfolio from commodity to value-added products.

• Journal of Korean Society of Forest Science
• /
• v.103 no.3
• /
• pp.375-382
• /
• 2014

The objective of this study was to investigate suitable plantation site for planted Chamaecyparis obtusa seedlings from the analysis of regional early growth performances. Two years old C. obtusa seedlings were planted with the density of $3,900seedlings{\cdot}ha^{-1}$ in late March, 2011 at Haeman and Jangseong, Jeollanamdo. In each study site, three plots ($400m^2$ per plot) were established and root collar diameter (mm) and tree height (cm) of each C. obtusa were measured in April, 2011 and October from 2011 to 2013. We also analyzed soil physical and chemical properties of sites and compartmental nitrogen and phosphorus concentrations of C. obtusa. Although the concentrations of soil nitrogen, organic matter, and C.E.C. at Haenam site were higher than those at Jangseong site, early growth performances of planted C. obtusa at Jangseong site were significantly better than those at Haenam site. The reasons for these results were probably related to deep available soil depth at Jangseong site and relatively low annual precipitation and sea wind at Haenam site, which was adjacent to the sea. The compartmental nitrogen and phosphorus concentrations of C. obtusa was in the order of needles > current twigs > fine root > stem and branches root above 2 mm in diameter. The phosphorus concentration of needles at Haenam site was significantly higher than that at Jangseong site. The results of this study might be useful for the selection of suitable plantation site for C. obtusa.

• Journal of Korea Technology Innovation Society
• /
• v.9 no.4
• /
• pp.842-864
• /
• 2006

This paper deals with an empirical study to statistically analyse various financial performances of the selected IPO firms using their investments on research and development(R&D) as an independent variables. The major results of statistical analyses have come up with the followings: 1) The regression analyses for change in average annual total market stock value/total assets over that of R&D expenditures showed the positive relationship, However, those of sales volume and net assets per share showed negative without statistical significances. 2) The statistical analyses in effect of the 3-year average total market stock value/total assets over the 3-year average R&D expenditures resulted in the positive coefficients what are statistically significant at 95% level. 3) Another statistical analysis showed that the financial performances of the IPO finns with deferred assets were better than those of the firms without them. In sum, the degree of investment on R&D by the IPO firms are expected to positively affect their financial performances except the Finns without having proper original technologies.

• Proceedings of the Safety Management and Science Conference
• /
• 2011.04a
• /
• pp.731-746
• /
• 2011

The successful and sustainable growth of SMEs depends on their ability of strengthen their competitiveness in quality and cost and service more than anything else as a fundamental of operation. Among these key competitive factors of SMEs, quality is the most critical factor in manufacturing business fields. Because quality strongly influence cost and service performance on this manufacturing business field. There are many different ways to improve the quality performance but it needs proper management decision to choose the best way what can maximize outputs with minimum inputs. And it needs effective measurement methods and some indicators to analysis the quality performance properly. The quality cost is one of the simplest key indicators to measure the quality performance and the effectiveness of quality related management decisions. The major purpose of this study is to diagnose the categorized current level of actual quality cost of local SMEs to maximize their quality management effectiveness through comparing their level with others what's expressed in early studies. In this study, through survey on local SMEs, we found that their average annual quality cost ratio versus turnover - Total amount of annual quality cost divided by annual turnover - is around 3.69% excluded some SME's performances what have different quality control measures with others. And we found some results what corresponded with the early studies on the correlations between those categorized quality costs factors and some discrepancies between some of the literature model and the early case study results as follows. There were negative correlations between the Prevention costs and the External failure costs, and the Appraisal costs and the External failure costs, and there was positive correlation between the Appraisal costs and Internal failure costs same as early studies. But, we couldn't found any strong negative correlations between the Cost of control - Preventive costs & Appraisal costs - and the Cost of Failure of control - Internal & External failure costs -. It reveals not only the lack of effectiveness on their preventive or appraisal activities but also it can reveal there were so many effective ways to prevent the failure costs properly such as some innovative investment on Factory automation includes Error Proofing and more preventive actions to improve the effectiveness of the typical management methods likes CE (Concurrent Engineering), APQP (Advanced Product Quality Planning), FMEA (Failure Mode & Effect Analysis) etc.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.56 no.4
• /
• pp.9-19
• /
• 2014

The objective of this study was to evaluate of the Tank model in simulating runoff discharge from rural watershed in comparison to the SWAT (Soil and Water Assessment Tool) model. The model parameters of SWAT was calibrated by the shuffled complex evolution-university Arizona (SCE-UA) method while Tank model was calibrated by genetic algorithm (GA) and validated. Four dam watersheds were selected as the study areas. Hydrological data of the Water Management Information System (WAMIS) and geological data were used as an input data for the model simulation. Runoff data were used for the model calibration and validation. The determination coefficient ($R^2$), root mean square error (RMSE), Nash-Sutcliffe efficiency index (NSE) were used to evaluate the model performances. The result indicated that both SWAT model and Tank model simulated runoff reasonably during calibration and validation period. For annual runoff, the Tank model tended to overestimate, especially for small runoff (< 0.2 mm) whereas SWAT model underestimate runoff as compared to observed data. The statistics indicated that the Tank model simulated runoff more accurately than the SWAT model. Therefore the Tank model could be a good tool for runoff simulation considering its ease of use.

• The Korean Journal of Health Service Management
• /
• v.10 no.2
• /
• pp.15-23
• /
• 2016

Objectives : The purpose of this study was to evaluate the productivity changes of 18 public health centers in Gangwon-do from 2006 to 2013 using the Malmquist Productivity Index(MPI). Methods : Data were collected from Statistics Korea from 2006 to 2013. The input variables were the numbers of medical, nursing and administrative personnels. The output variables were the performances of health promotion programs. Along with the traditional input-oriented DEA analysis, the MPI was calculated. Results : First, among the 18 public health centers, the productivity index of 14 public health centers was increased. Second, the annual productivity showed a 6% increase. Third, the productivity improvements were mainly caused by Scale Efficiency Change. Conclusions : Improving the productivity of public health centers requires the support and external policies of the national and local government. Internally, public health centers need to maintain scale optimization of the center. Additionally, efforts should be made to effectively use limited resources.

• Korean Journal of Air-Conditioning and Refrigeration Engineering
• /
• v.23 no.6
• /
• pp.392-399
• /
• 2011

Ground-coupled heat pump(GCHP) systems have been shown to be an environmentally-friendly, efficient alternative to traditional cooling and heating systems in both residential and commercial applications. Although some work related to performance evaluation of GCHP systems for commercial buildings has been done, relatively little has been reported on the residential applications. The aim of this study is to evaluate the cooling and heating performances of a vertical GCHP system applied to an artificial detached house($117\;m^2$) in Seoul. For this purpose, a typical design procedure was involved with a combination of design parameters such as building loads, heat pump capacity, borehole diameter, and ground thermal properties, etc. The cooling and heating performance simulation of the system was conducted with different prediction times of 8760 hours and 240 months. The performance characteristics including seasonal system COP, average annual power consumption, and temperature variations related to ground heat exchanger were calculated and compared.