• Journal of Information Processing Systems
• /
• v.14 no.1
• /
• pp.79-100
• /
• 2018

Twitter is one of the most popular social platforms for online users to share trendy information and views on any event. Twitter reports an event faster than any other medium and contains enormous information and views regarding an event. Consequently, Twitter topic summarization is one of the most convenient ways to get instant gist of any event. However, the information shared on Twitter is often full of nonstandard abbreviations, acronyms, out of vocabulary (OOV) words and with grammatical mistakes which create challenges to find reliable and useful information related to any event. Undoubtedly, Twitter event summarization is a challenging task where traditional text summarization methods do not work well. In last decade, various research works introduced different approaches for automatic Twitter topic summarization. The main aim of this survey work is to make a broad overview of promising summarization approaches on a Twitter topic. We also focus on automatic evaluation of summarization techniques by surveying recent evaluation methodologies. At the end of the survey, we emphasize on both current and future research challenges in this domain through a level of depth analysis of the most recent summarization approaches.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2021.07a
• /
• pp.395-398
• /
• 2021

최근 방대한 양의 텍스트 정보가 인터넷에 유통되면서 정보의 핵심 내용을 파악하기가 더욱 어려워졌으며, 이로 인해 자동으로 텍스트를 요약하려는 연구가 활발하게 이루어지고 있다. 텍스트 자동 요약을 위한 다양한 기법 중 특히 트랜스포머(Transformer) 기반의 모델은 추상 요약(Abstractive Summarization) 과제에서 매우 우수한 성능을 보이며, 해당 분야의 SOTA(State of the Art)를 달성하고 있다. 하지만 트랜스포머 모델은 매우 많은 수의 매개변수들(Parameters)로 구성되어 있어서, 충분한 양의 데이터가 확보되지 않으면 이들 매개변수에 대한 충분한 학습이 이루어지지 않아서 양질의 요약문을 생성하기 어렵다는 한계를 갖는다. 이러한 한계를 극복하기 위해 본 연구는 소량의 데이터가 주어진 환경에서도 양질의 요약문을 생성할 수 있는 문서 요약 방법론을 제안한다. 구체적으로 제안 방법론은 한국어 사전학습 언어 모델인 KoBERT의 임베딩 행렬을 트랜스포머 모델에 적용하는 방식으로 문서 요약을 수행하며, 제안 방법론의 우수성은 Dacon 한국어 문서 생성 요약 데이터셋에 대한 실험을 통해 ROUGE 지표를 기준으로 평가하였다.

• Annual Conference on Human and Language Technology
• /
• 2020.10a
• /
• pp.167-171
• /
• 2020

문서 요약은 길이가 긴 원본 문서에서 의미를 유지한 채 짧은 문서나 문장을 얻어내는 작업을 의미한다. 딥러닝을 이용한 자연어처리 기술들이 연구됨에 따라 end-to-end 방식의 자연어 생성 모델인 sequence-to-sequence 모델을 문서 요약 생성에 적용하는 방법들이 연구되었다. 본 논문에서는 여러 자연어처리 분야에서 높은 성능을 보이고 있는 BERT 모델을 이용한 자연어 생성 모델에 복사 메커니즘과 강화 학습을 추가한 문서 요약 모델을 제안한다. 복사 메커니즘은 입력 문장의 단어들을 출력 문장에 복사하는 기술로 학습데이터에서 학습되기 힘든 고유 명사 등의 단어들에 대한 성능을 높이는 방법이다. 강화 학습은 정답 단어의 확률을 높이기 위해 학습하는 지도 학습 방법과는 달리 연속적인 단어 생성으로 얻어진 전체 문장의 보상 점수를 높이는 방향으로 학습하여 생성되는 단어 자체보다는 최종 생성된 문장이 더 중요한 자연어 생성 문제에 효과적일 수 있다. 실험결과 기존의 BERT 생성 모델 보다 복사 메커니즘과 강화 학습을 적용한 모델의 Rouge score가 더 높음을 확인 하였다.

• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.139-143
• /
• 2021

최근 대용량 말뭉치를 기반으로 한 언어 모델이 개발됨에 따라 다양한 자연어처리 분야에서 사람보다 높은 성능을 보이는 시스템이 제안되었다. 이에 따라, 더 어렵고 복잡한 문제를 해결하기 위한 데이터셋들이 공개되었으며 대표적으로 기계독해 작업에서는 시스템이 질문에 대해 답변할 수 없다고 판단할 수 있는지 평가하기 위한 데이터셋이 공개되었다. 입력 받은 데이터에 대해 답변할 수 없다고 판단하는 것은 실제 애플리케이션에서 중요한 문제이기 때문에, 이를 해결하기 위한 연구도 다양하게 진행되었다. 본 논문에서는 문서를 이해하여 답변할 수 없는 데이터에 대해 효과적으로 판단할 수 있는 기계독해 시스템을 제안한다. 제안 모델은 문서의 내용과 질문에 대한 이해도가 낮을 경우 정확한 정답을 맞히지 못하는 사람의 독해 패턴에서 착안하여 기계독해 시스템의 문서 이해도를 높이고자 한다. KLUE-MRC 개발 데이터를 통한 실험에서 EM, Rouge-w 기준으로 각각 71.73%, 76.80%을 보였다.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.382-387
• /
• 2023

현재 GPT-4와 같은 거대한 언어 모델이 기계 번역, 요약 및 대화와 같은 다양한 작업에서 압도적인 성능을 보이고 있다. 그러나 이러한 거대 언어 모델은 학습 및 적용에 상당한 계산 리소스와 도메인 특화 미세 조정이 어려운 등 몇 가지 문제를 가지고 있다. In-Context learning은 데이터셋에서 추출한 컨택스트의 정보만으로 효과적으로 작동할 수 있는 효율성을 제공하여 앞선 문제를 일부 해결했지만, 컨텍스트의 샷 개수와 순서에 민감한 문제가 존재한다. 이러한 도전 과제를 해결하기 위해, 우리는 Super In-Context Learning (SuperICL)을 활용한 새로운 방법론을 제안한다. 기존의 SuperICL은 적용한 플러그인 모델의 출력 정보를 이용하여 문맥을 새로 구성하고 이를 활용하여 거대 언어 모델이 더욱 잘 분류할 수 있도록 한다. Super In-Context Learning for Generation은 다양한 자연어 생성 작업에 효과적으로 최적화하는 방법을 제공한다. 실험을 통해 플러그인 모델을 교체하여 다양한 작업에 적응하는 가능성을 확인하고, 자연어 생성 작업에서 우수한 성능을 보여준다. BLEU 및 ROUGE 메트릭을 포함한 평가 결과에서도 성능 향상을 보여주며, 선호도 평가를 통해 모델의 효과성을 확인했다.

• Journal of the Korean Chemical Society
• /
• v.55 no.3
• /
• pp.400-404
• /
• 2011

Safflower is called as the 'beneficial flower' because 'it helps human health', and it was introduced as red flower in Tonguibogam due to the red color of floral leaf. From old times, it has been used for the material of cloth and rouge. Recently, polyphenol compound, the main ingredient of safflower, known as anti-aging and anti-oxidizing material in the healthy food industry becomes the emerging hot topic. This study aims to confirm by DDT (Disk Diffusion Test) assay, MTT assay, and NF-${\kappa}$B Luciferase activity inhibition assay in vitro that polyphenol compound, which is the main ingredient of safflower, has the anti-microbial efficacy to inhibit the growth of acne germs that make troubles for the teenagers or middle aged. Also it aims to evaluate its clinical efficacy on the acne skin, utilizing the facial cleansing cosmetic form of soap sample. This study can contribute to take a major step forward to the development of cosmetic soap for acne in the cosmeceutical industry.

• Journal of Environmental Science International
• /
• v.5 no.5
• /
• pp.585-592
• /
• 1996

Trifluralin (2, 6-dinitro-N, N-dipropyl-4- (trifluormethyl) benzenamine) was applied preemergent to soybean in plots drained or nondrained, in louisiana. Plots 14.6 ha were arranged to give 1683 g/ha of trifluralin. The half life of trifluralin in the top 15 cm of soil was 42.6 darts and f6.0 days in nondrained plot and drained plot, respectively. The concentrations of trifluralin in surface runoff water and subsurface runoff water were 0.62 ng/ml-0.02 ng/ml and 11.06 ng/ml-0.02 ng/ml, respectively. The concentration of trifluralin in runoff water was smaller than 2 ng/ml for trifluralin of U.S. Environmental Protection Agency advisory. Total loss of trifluralin in runoff water was 0.021 % of applied amount at drained plots during three month after application. Trifluralin was moved hardly in the water. Subsurface drainage -reduced trifluralin losses because concentration of trifluralin in the subsurface runoff water in drained fields was low.

• Journal of the Korea Fashion and Costume Design Association
• /
• v.12 no.4
• /
• pp.115-135
• /
• 2010

Regarding the figure painting of the T'ang period, by Yeomipbon, a representative painter of the early T'ang period, shows a traditional painting style. Figure paintings of the Prosperous T'ang period, influenced by Odoja's painting style, are characterized as illustrating the lives of noble women delicately with brilliant colors. They include Janghwon's and in which Chinese traditional and Western painting styles were combined. , and , which were painted by Jubang in the mid T'ang period, were the developmentally succeeded the figure painting style of the previous times, illustrating even the psychological and emotional states of painted figures in a sophisticated way. Skin make-ups shown in figure paintings of the T'ang period are mostly baekjang in type. Besides, other different types of the make=up such as bihajang, dohwajang and juwoonjang can be seen. The paintings, show that the tone of skin got more and more red as the times moved from the early to the late T'ang periods because rouge was more and more used as the times passed between the periods. As a type of forehead make-up, aekhwang is found in Janghwon's and of the late T'ang period. Hwajeon is seen in lots of figure paintings made between the Prosperous the late T'ang periods. Most eyelid make-ups shown in the figure paintings are round-shaped wolmi and yuhyeopmi in type. Besides, other creative types like gyeyeopmi, paljami and iljami are found. Lips are found made up clearer and more red as the times went from the early and the late T'ang periods. The types of cheek make-up like sahong and jangyeop can't be found in figure paintings of the T'ang period.

• Proceedings of the Korean Geotechical Society Conference
• /
• 2010.09b
• /
• pp.152-160
• /
• 2010

In this study, cohesion of soft ground, soft ground depth and embankment height varying conditions, such as the impact of each condition after the calculation of the range, SCP was performed to evaluate the applicability of the method. Reinforcing effects of scope, and permit lateral movement of SCP 2D and 3D analysis of the program were calculated by the displacement ratio, the result follows. The height and depth of soft soil embankment with increasing and decreasing the cohesion tends to be affected were long range, SCP method applied by the finite element analysis Cu = 1.0tf/$m^2$, embankment height is 3.0m depth of soft soil can be applied in a less than 5.0m, and Cu = 3.0tf/$m^2$, embankment height, the soft soil depth is 3.0m 12.0m, Cu = 3.0tf/$m^2$, embankment height is 5.0m less than 7.0m depth of soft soil can be applied in was. And Cu = 5.0tf/$m^2$, embankment height is 3.0m below 15.0m depth rouge anti Floor, Cu = 3.0tf/$m^2$, embankment height of 5.0m 12.0m depth below the soft soil, Cu = 5.0tf/$m^2$, If the depth of soft soil embankment height of 7.0m and below 5.0m was applicable.

• Journal of Environmental Science International
• /
• v.4 no.4
• /
• pp.379-386
• /
• 1995

Permethrin [3-phenox yben zyl(1RS)-cis,tans-3-(2,2-dichloroviny1)-2,2- dimethylcyclopropane carboxylate] insecticides were selected to study adsorption and leaching potentials related to pollution on Commerce silty clay loam soil near Baton Rouge, Louisiana, USA. GLC-ECD chromatogram of permethrin included 32.5 % of cis-pemehin and 67.4 % of trans-permethrin. Extraction efficiencies of is and trmu-pemetun were 92.5 % and 92.3 % in fortified water, respectively and 85.9 % and 88.8 % in fortified soil, respectively. At a 1:10 moi]/water ratio,the Koc values for sis and fan isomers of permethrin were 938 and 877, respectively. Leaching of permethrin was evaluated in soil columns$(5.4 cm i.d. \times 26 cm length)$. Total recoveries of the permethrin applied to the soil column were $84.5\pm3.1%$. When the soil columns were leached with three pore volumes of water, the distributions of cis-permethrin leached were 6.10 % and 0.07 % of amount applied in the untreated zone soil and leachate water, respectively Trans-pemethrin distributions were 5.20 % in the untreated zone soil and 0.05 % in leachate water. Cis and trans-pemethrin was strongly adsorbed to soil. The results of the study showed the strong relationship between adsorption and leaching. Cis and trans-permethrin to be leached into the groundwater in soils with shallow aquifers were suggested a low leaching potential.