• Annual Conference of KIPS
• /
• 2023.11a
• /
• pp.493-494
• /
• 2023

음성 기반의 감정 분석은 인간의 감정을 정확하게 파악하는 데 중요한 연구 분야로 자리잡고 있다. 최근에는 wav2vec2.0과 같은 트랜스포머 기반의 모델이 음성 인식 분야에서 뛰어난 성능을 보이며 주목받고 있다. 본 연구에서는 wav2vec2.0 모델을 활용하여 한국어 감성 발화 데이터에 대한 감정 분류를 위한 데이터 샘플링 전략을 제안한다. 실험을 통해 한국어 음성 감성분석을 위해 학습 데이터를 활용할 때 감정별로 샘플링하여 데이터의 개수를 유사하게 하는 것이 성능 향상에 도움이 되며, 긴 음성 데이터부터 이용하는 것이 성능 향상에 도움이 됨을 보인다.

• Journal of Information Technology Applications and Management
• /
• v.30 no.3
• /
• pp.15-35
• /
• 2023

Considerable agreement exists about the importance of promoting entrepreneurship in both developed and developing countries. In less developed countries, governments see entrepreneurship as a way to stimulate economic development and tackle serious economic and social challenges. So how can countries encourage young people to become entrepreneurs? Research confirms that intentions play an important role in the decision to start a new firm and many factors influence that intentions. The purpose of the study was to investigate the antecedents to entrepreneurship behavior with particular attention to moderating role of social support and entrepreneur self-efficacy. The study covered 116 business students of undergraduate and post graduate level studying under different universities in Kathmandu, Nepal. The questionnaire for data collection was distributed in college groups via WhatsApp and viber with the support and permission from the college administration. The study design used was correlational with a sampling procedure of convenience. The study only showed the impact of attitude to entrepreneurship behavior as well as moderating effect of social support was also observed.

• Journal of Information Technology Applications and Management
• /
• v.30 no.5
• /
• pp.59-82
• /
• 2023

The present research aims to examine whether learning from the supervisor influences readiness for change with the mediating impact of planfulness. Drawing upon the theory of planned behavior, it is hypothesized that learning from the supervisor positively impacts planfulness ability in individuals, which in turn enhances the readiness for change. Through using convenience sampling, the sample of 451 was collected from employees working full-time in the manufacturing and I.T. service organizations in India. Structural equation modeling and regression analysis indicate that learning from the supervisor is positively associated with readiness for change and planfulness. Additionally, planfulness fully mediated the relationship between learning from the supervisor and readiness to change. The findings of the present research highlight that continuous support and learning from the supervisor enhances the planfulness ability of the individual and consequently enhances individual readiness for change. The current research is pioneering in testing the hypothetical model associating learning from the supervisor, planfulness, and readiness for change.

• Journal of Information Technology Applications and Management
• /
• v.28 no.5
• /
• pp.51-60
• /
• 2021

The main purpose of this study is to examine the factors that influence the employee's intention to leave the current job. A descriptive survey design was conducted and primary data were collected by using questionnaire. The study was based on random sampling; a sample of 200 was drawn from the employees who work in different micro finance institutions in Nepal. For the data analysis percentage analysis and chi square-test were used. The study revealed that salary, working condition, career growth, supervisor behavior and outstation influence the job quitting process. The study also defined the association between the reasons for leaving job such as salary, working condition, career growth supervisor behaviour & outstation and satisfaction, time spent in the organisation, training & development facility, frequency of changing job & encourage from immediate boss. This study establishes the importance of intentions of employees to quit job and provides further evidence in revealing the intentions to quit among the employees.

• Horticultural Science & Technology
• /
• v.32 no.1
• /
• pp.84-90
• /
• 2014

Berchemia racemosa var. magna is only found in Anmyeon Island of South Korea. Genetic diversity and the spatial genetic structure of B. racemosa var. magna in Anmyeon Island were studied by I-SSR marker system. Fifty I-SSR amplicons were produced from 8 selected primers. We used 13 polymorphic markers to analyze the genetic structure. Distribution of 39 individuals in the study plot($90m{\times}70m$) showed aggregate pattern (aggregation index = 0.706). Total 21 genets were observed from 39 individuals through I-SSR genotyping. Proportion of distinguishable genotype (G/N), genotype diversity (D) and genotype evenness (E) were 53.8%, 0.966 and 0.946, respectively. In spite of the small number and the narrow distribution, Shannon's diversity index (I = 0.598) was relatively high as compared with those of the other plant species. For ex situ genetic conservation of B. racemosa var. magna, the sampling strategy based on spatial autocorrelation using Tanimoto distance is efficient at choosing the conserved individuals with a 6 meter interval between individual trees.

• Korean Security Journal
• /
• no.12
• /
• pp.177-200
• /
• 2006

This research result that analysis as following. First, difference that keep in mind partially as result that verify difference of job-stress by demography special qualitys that is common people guard's society and turnover intention appeared. Second, result that analyze relation with job-stress and turnover intention, displayed statical correlation in job environment. Third, result that keep in mind as statistical as result that verify effect that common people guard's job-stress gets to turnover intention appeared. During is job-stress variable, was expose that exert effect that keep in mind to turnover intention job surrounding and job special quality. Suggestion point through this research as following be. To raise common people guard's job-stress and turnover intention. system improvement strategy about social estimation etc.. is effective, and do on the basis of this and may have to establish common people guard's enhancing morale plan strategy, low complement that can exert negative impact in change of post for example, line of duty that it is seldom competence, daytime that can elicit common people guards' job satisfaction. Is in visual point that activation of continuous research is leaned against by depth interview and review through qualitative research was insufficient relatively in virtue research so far in research after.

• Korean Journal of Ecology and Environment
• /
• v.34 no.3 s.95
• /
• pp.223-238
• /
• 2001

Longitudinal distribution of the fish fauna of the 20 main channel and tributary sites in the Nakdong River was investigated from July 1999 to December 2000. A total of 4,791 individuals of fisHes were collected from three to seven sampling programs of all study sites and 48 species of 17 families were collected. Dominant family was Cyprinidae (25 species, relative abundance (RA): 91.8%). Zacco platypus (RA: 18.0%) was found to be the most abundant inhabitant. Subdominant species were Opsariichthys bidens (17,1%), Squalidus chankaensis tsuchigae (13.2%), Hemibarbus labeo (9.9%), and Z. temmincki (8.2%). Among the total species, 12 species were identified as endemic species (4 families, 24.5% of 1,172 individuals). Four exotic fishes were collected (Carassius cuvieri, 25 indv.; Micropterus salmoides, 214 indv.; Lipomis macrochirus, 14 indv.; Oreochromis niloticus, 6 indv.). M. salmoides was distributed in mid to lower part of the Nakdong River. As the result of cluster analysis, sampling sites were divided into three groups (upper, middle and lowerpart). Site below Gumi (St. 10) were considered as the lower part of the Nakdong River. Combined with previous surveys, about 80 species are present in the main channel and tributaries of the Nakdong River. Since the middle of 1990s, population density of the M. salmoides rapidly increased in the stagnant lower part of the river(RA 8.8% in lower 5 sites). The feeding strategy of M. salmoides was "generalized" on Daphnia sp. (<100 mm), and "specialization" about Cladocera (<150 mm) and Insecta(<200 mm).

• Korean Journal of Plant Resources
• /
• v.25 no.4
• /
• pp.407-415
• /
• 2012

Pinus pumila, which occurs in the northeast Asia, is found limitedly in Daecheongbong area of Mt. Seorak in the South Korea. This population was chosen to study spatial pattern, genetic diversity and spatial genetic structure. There were 48 polymorphic and 30 monomorphic I-SSR markers. A total of 65 individuals which distributed in the study site (40 m ${\times}$ 70 m) showed weakly aggregate distribution (Aggregate Index = 0.871). A total of 40 genets were observed from 65 individuals through I-SSR genotype comparison. Proportion of distinguishable genotype (G/N), genotype diversity (D) and genotype evenness (E) were 61.5%, 0.977 and 0.909, respectively. In spite of the small number and the limited distribution, Shannon's diversity index (I = 0.567) was relatively high as compared with those of other plant species. Spatial autocorrelation using Tanimoto's distance showed that the genetic patch was established within 12 m. Based on Mantel tests, there was relatively low correlation between genetic distance and geographic distance. Therefore, it seems the P. pumila population was formed by many parent trees in early stage. For ex situ genetic conservation of P. pumila, the sampling strategy is efficient at least above 12 m between individual trees.

• Journal of Korean Society of Occupational and Environmental Hygiene
• /
• v.21 no.1
• /
• pp.11-24
• /
• 2011

Working in a research laboratory means exposure to a wide range of hazardous substances. Several studies indicated that laboratory workers, especially working with chemicals, might have an increased risk of certain cancers. However, exposure assessment data in laboratory settings are scarce. This study was performed to examine several approaches for quantitatively assessing the exposure levels to volatile organic compounds (VOCs) among workers in chemistry laboratories. The list of 10 target VOCs, including ethanol, acetone, 2-propanol, dichlormethane, tetrahydrofuran, benzene, toluene, n-hexane, ethyl acetate, chloroform, was determined through self-administered questionnaire for six chemistry research laboratories in a university, a government-funded research institute, or private labs. From September to December 2008, 84 air samples were collected (15 area samples, 27 personal time weighted samples, 42 personal task-basis short-term samples). Real time monitors with photo ionization detector were placed during the sampling periods. In this study, benzene was observed exceeding the action levels, although all the results were below the American Conference of Governmental Industrial Hygienists (ACGIH) Threshold Limit Value (TLV). From the air sampling results, we concluded that (1) chemicals emitted during experiments could directly affect to neighbor office areas (2) chemical exposures in research laboratories showed a wide range of concentrations depending on research activities (3) area samples tended to underestimate the exposures relative to personal samples. Still, further investigation, is necessary for developing exposure assessment strategies specific to laboratories with unique exposure profiles.

• The Korean Journal of Applied Statistics
• /
• v.24 no.3
• /
• pp.477-484
• /
• 2011

When an event of target variable is rare, a widespread strategy is to build a model on the sample that disproportionally over-represents the events, that is over-sampled. Using the data over-sampled from the original data set, the predicted values would be biased; however, it can be easily corrected to represent the population. In this study, we investigate into the relationship between the proportion of rare event on a data-mart and the model performance using real world data of a Korean credit card company. Also, we use the methods for adjusting of posterior probability for over-sampled data of the offset method and the weighted method. Finally, we compare the performance of the methods using real data sets.