• Journal of the Korean Society of Food Science and Nutrition
• /
• v.39 no.8
• /
• pp.1126-1131
• /
• 2010

The antiobesity effect of soymilks fermented with Bacillus subtilis KC-3 (KCCM 42923) from cheonggukjang was compared with other sources of B. subtilis KCCM 11316 and B. subtilis MYCO. The antiobesity effect was investigated by measuring the release of leptin, Oil red O staining, glycerol secretions and adipogenic transcription factor by reverse transcription-polymerase chain reaction (RT-PCR) in the 3T3-L1 adipocytes. Fermented soymilk with B. subtilis KC-3 (F-KC) led to decrease levels of leptin secretion and increase levels of glycerol secretion in the cells. In addition, F-KC reduced contents of Oil red O dye in the 3T3-L1 adipocytes. Also, mRNA expression levels of both SREBP-1c (sterol regulatory element-binding protein 1-c) and PPAR-$\gamma$ (peroxisome proliferator-activated receptor-$\gamma$), which are adipogenic transcription factor, in cells treated with F-KC were markedly down regulated. These results demonstrate that the Bacillus subtillis fermented soymilk (F-KC) decreased lipid content in 3T3-L1 adipocytes by inhibiting lipogenesis. All B. subtilis fermented soymilks had shown antiobesity activities, however, F-KC exhibited the strongest antiobesity effect in the 3T3-L1 adipocytes. Our study suggests that especially F-KC increased the potential of antiobesity effects.

• Korean Journal of Environmental Agriculture
• /
• v.11 no.2
• /
• pp.155-162
• /
• 1992

In order to extend the releasing period of granular formulation to approximately 20 days, the KC-6620-adsorbed granules were formulated with carriers and polyethylene glycol as adjuvant. The releasing rates of active ingredient from the formulations were evaluated in aqueous medium. The baked bentonite was found most effective carrier to sustain the release of KC-6620. Due to, however, low releasing rate of active ingredient after 20 days, bentonite formulation appeared to be of no practical for the short-term sustained release of KC-6620. The increased pore volume of bentonite granular formulation by adding pyrophyllite increased remarkably the released amount of KC-6620 from bentonite-pyrophyllite(4 : 6) granule up to 85% of total active ingredient incorporated. Addition of polyethylene glycol to the bentonite-pyrophyllite granule further increased the releasing rate of KC-6620. With KC-6620 content in the bentonite-pyrophyllite(4 : 6) granule, the releasing rate of active ingredient was markedly reduced.

• Journal of Korean Society of Disaster and Security
• /
• v.11 no.2
• /
• pp.61-67
• /
• 2018

Scour protection methods can be categorized as two types: The first is to reduce the horseshoe and wake vortices which are the main reasons for local scour. Either small cylindrical structures or separated vertical deflectors can be placed in front of the pier or the horizontal deflector (or collar) can be attached to the pier like the spoiler to reduce the dynamics of vortical structures. The second is to employ the protection layer to keep the bed material in place, which is a common method with a merit of immediate effect by using block mat or tetrapod. This study examined the effect of scour reduction using the former method. The relationship between the reflector interval and reduction of scour was not clear. It is assumed that the width of the reflector is somewhat correlated with the reduction of the scour. As the KC numbers increases, the Effect of Scour reduction rate is shown to decrease. Also, Scour reduction rate showed a rapid change at $U_R=25$ or KC = 8.

• Current Research on Agriculture and Life Sciences
• /
• v.20
• /
• pp.9-18
• /
• 2002

A total of 31 C. chinense lines selected from 2000 screening were tested again for resistance to P. capsid but resistance was not found in tile lines. A total of 26 selections from Korean land races tested 2001 spring were tested again for resistance to P. capsici, KC180, KC230, KC195 and KC194 showed moderate resistance to P. capsid. However, it was apparent on the basis of horticultural characteristics that KC180 and KC230 had been naturally crossed with AC2258 and CM334, respectively. CM334 and AC2248 seed lots that were increased in different years were taken out and tested for resistance to improve their genetic purity because the resistant genetic resources have been showing some off-types in tile population. Off-types began to be found in 1992 seed lots and tile proportion and degree of tile offs was increasing with time up to 2001. Plants true to the type in 1992 seed lots were selected and their inbred seeds were mass produced in a net cage in the greenhouse. AC2258 included in the experiment together was uniform. In 1995 seed lots of CM334, plants with resistance to P. capsici and low or no number of lateral branching at cotyledonary axil, although they were off from tile original CM334, were found and selection was applied to breed lines fixed in tile characters.

• Korean Journal of Cognitive Science
• /
• v.14 no.4
• /
• pp.21-31
• /
• 2003

This paper reports our experimental study with Korean-Chinese (=KC) bilinguals as compared with Korean monolinguals. We aim to find KC bilingual speakers' sentence processing strategies, and the interaction between the Ll and U2 grammars in bilingual development. To this end, we recruited 166 subjects of all age groups from age 3 to adult in the Korean autonomous community in Yanji, China, and did a classical subject/actor identification test, where subjects are supposed to pick out the subjects/actor of both sensical and nonsensical sentences (cf. Liu, Bates & Li, 1992). We compared our results with our previous work on monolingual Koreans, and found out that KC bilinguals rely on word order as well as anumacy; that KC bilinguals make use of morphology at age 10 as compared with age 5 for monolinguals; and that KC bilingual adults rely on animacy and word order as well as morphology, while monolingual Korean adults rely solely on morphology for sentence interpretation. Given that animacy and word order play an important role in the Chinese grammar, our finding lends support to the backward transfer which Liu, Bates & Li (1992) propose for early bilingualism.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.577-580
• /
• 2023

악성댓글은 인터넷상에서 정서적, 심리적 피해를 주는 문제로 인식되어 왔다. 본 연구는 한국어 악성댓글 탐지 분석을 위해 KcBERT 및 다양한 모델을 활용하여 성능을 비교하였다. 또한, 공개된 한국어 악성댓글 데이터가 부족한 것을 해소하기 위해 기계 번역을 이용하고, 다국어 언어 모델(Multilingual Model) mBERT를 활용하였다. 다양한 실험을 통해 KcBERT를 미세 조정한 모델의 정확도 및 F1-score가 타 모델에 비해 의미 있는 결과임을 확인할 수 있었다.

• Annual Conference on Human and Language Technology
• /
• 2020.10a
• /
• pp.437-440
• /
• 2020

최근 자연어 처리에서는 사전 학습과 전이 학습을 통하여 다양한 과제에 높은 성능 향상을 성취하고 있다. 사전 학습의 대표적 모델로 구글의 BERT가 있으며, 구글에서 제공한 다국어 모델을 포함해 한국의 여러 연구기관과 기업에서 한국어 데이터셋으로 학습한 BERT 모델을 제공하고 있다. 하지만 이런 BERT 모델들은 사전 학습에 사용한 말뭉치의 특성에 따라 이후 전이 학습에서의 성능 차이가 발생한다. 본 연구에서는 소셜미디어에서 나타나는 구어체와 신조어, 특수문자, 이모지 등 일반 사용자들의 문장에 보다 유연하게 대응할 수 있는 한국어 뉴스 댓글 데이터를 통해 학습한 KcBERT를 소개한다. 본 모델은 최소한의 데이터 정제 이후 BERT WordPiece 토크나이저를 학습하고, BERT Base 모델과 BERT Large 모델을 모두 학습하였다. 또한, 학습된 모델을 HuggingFace Model Hub에 공개하였다. KcBERT를 기반으로 전이 학습을 통해 한국어 데이터셋에 적용한 성능을 비교한 결과, 한국어 영화 리뷰 코퍼스(NSMC)에서 최고 성능의 스코어를 얻을 수 있었으며, 여타 데이터셋에서는 기존 한국어 BERT 모델과 비슷한 수준의 성능을 보였다.

• Journal of the Korean Society for Marine Environment & Energy
• /
• v.6 no.1
• /
• pp.60-67
• /
• 2003

Flow around an oscillating circular cylinder was experimentally investigated. With varying Keulegen-Carpenter(KC) number from 10 to 30 the flow field with vortex sheddings and the related hydrodynamic fortes exerting on the cylinder were measured. A newly developed PW(paricle image velocimetry) successfully captured the complex vortical flows varying with the KC number and the flow patterns were 'traverse street', 'single pairing' and 'double pairing' of vortices with increasing KC number, At a certain KC number range the lift force undergoes a transition showing little periodicity due to surrounding complicated shedded vortices.

• Korean Journal of Weed Science
• /
• v.14 no.2
• /
• pp.81-93
• /
• 1994

This study was conducted to characterize herbicidal activity of the new type diphenylether compounds inducing bleaching(whitening) in relation to their chemical structures and to find out the herbicidal action mechanism. Bleaching was highly appeared in the compounds which have short chain alkyl or allyl group of cabamoyl type in meta position of A ring and nitro group in para position of B ring, and KC6361 showed the highest efficacy among these derivatives. Rice was tolerant to KC6361 at 2.0kg/ha. The applications of this compound at 0.25~1.0kg/ha provided acceptable control of the various weeds including large crabgrass, barnyardgrass, pigweed and so on. In the dark condition, KC6361 decreased carotenoid contents but increased chlorophyll in etiolated cucumber cotyledon. By the increase of light intensity, the carotenoid biosynthesis was more inhibited than the chlorophyll biosynthesis. The incorporation of $C^{14}$-acetate into lipid was slightly inhibited. Both KC6361 and norflurazon stimulated growth of dwarf rice(cv. Daneunbanju). These results suggest that both KC6361 and norflurazon may have the same action sites. To confirm this hypothesis, the changes of carotenoid intermediates in cucumber cotyledon and barnyardgrass leaf treated with KC6361 were investigated. Phytoene and phytofluene were increased but ${\beta}$-carotene was decreased, indicating that KC6361 inhibited phytoene and/or phytofluene dehydrogenase like norflurazon.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.4
• /
• pp.174-180
• /
• 2024

In this paper, we explore an emotion classification method through multimodal learning utilizing wav2vec 2.0 and KcELECTRA models. It is known that multimodal learning, which leverages both speech and text data, can significantly enhance emotion classification performance compared to methods that solely rely on speech data. Our study conducts a comparative analysis of BERT and its derivative models, known for their superior performance in the field of natural language processing, to select the optimal model for effective feature extraction from text data for use as the text processing model. The results confirm that the KcELECTRA model exhibits outstanding performance in emotion classification tasks. Furthermore, experiments using datasets made available by AI-Hub demonstrate that the inclusion of text data enables achieving superior performance with less data than when using speech data alone. The experiments show that the use of the KcELECTRA model achieved the highest accuracy of 96.57%. This indicates that multimodal learning can offer meaningful performance improvements in complex natural language processing tasks such as emotion classification.