• Title/Summary/Keyword: 지시자

Search Result 422, Processing Time 0.028 seconds

A Comparative Study on Korean Zero-shot Relation Extraction using a Large Language Model (거대 언어 모델을 활용한 한국어 제로샷 관계 추출 비교 연구)

  • Jinsung Kim;Gyeongmin Kim;Kinam Park;Heuiseok Lim
    • Annual Conference on Human and Language Technology
    • /
    • 2023.10a
    • /
    • pp.648-653
    • /
    • 2023
  • 관계 추출 태스크는 주어진 텍스트로부터 두 개체 간의 적절한 관계를 추론하는 작업이며, 지식 베이스 구축 및 질의응답과 같은 응용 태스크의 기반이 된다. 최근 자연어처리 분야 전반에서 생성형 거대 언어모델의 내재 지식을 활용하여 뛰어난 성능을 성취하면서, 대표적인 정보 추출 태스크인 관계 추출에서 역시 이를 적극적으로 활용 가능한 방안에 대한 탐구가 필요하다. 특히, 실 세계의 추론 환경과의 유사성에서 기인하는 저자원 특히, 제로샷 환경에서의 관계 추출 연구의 중요성에 기반하여, 효과적인 프롬프팅 기법의 적용이 유의미함을 많은 기존 연구에서 증명해왔다. 따라서, 본 연구는 한국어 관계 추출 분야에서 거대 언어모델에 다각적인 프롬프팅 기법을 활용하여 제로샷 환경에서의 추론에 관한 비교 연구를 진행함으로써, 추후 한국어 관계 추출을 위한 최적의 거대 언어모델 프롬프팅 기법 심화 연구의 기반을 제공하고자 한다. 특히, 상식 추론 등의 도전적인 타 태스크에서 큰 성능 개선을 보인 사고의 연쇄(Chain-of-Thought) 및 자가 개선(Self-Refine)을 포함한 세 가지 프롬프팅 기법을 한국어 관계 추출에 도입하여 양적/질적으로 비교 분석을 제공한다. 실험 결과에 따르면, 사고의 연쇄 및 자가 개선 기법 보다 일반적인 태스크 지시 등이 포함된 프롬프팅이 정량적으로 가장 좋은 제로샷 성능을 보인다. 그러나, 이는 두 방법의 한계를 지적하는 것이 아닌, 한국어 관계 추출 태스크에의 최적화의 필요성을 암시한다고 해석 가능하며, 추후 이러한 방법론들을 발전시키는 여러 실험적 연구에 의해 개선될 것으로 판단된다.

  • PDF

Lead Transport in Groundwater in Door County, Wisconsin (위스컨신주 도어지역의 지하수내 납성분의 이동)

  • Woo, Nam C.
    • Economic and Environmental Geology
    • /
    • v.27 no.1
    • /
    • pp.93-100
    • /
    • 1994
  • This study examined the mechanisms of lead transport in the groundwater system and those of irregular detections in groundwater in Door County, Wisconsin. During the spring recharge period in 1991, water-level movement and water-quality change were monitored from two monitoring wells equipped with three piezometers each and from five house wells, respectively. Water-level responses to recharge events were fast with a relatively short lag time ranging from 3 to 10 days, indicating that recharge of groundwater occurs through the high hydraulic conductivity (K) zones in the Silurian dolomite aquifer system. Lead was detected only on particles filtered from groundwater, but not in dissolved state. Concentrations ranged from 0.2 to $7.1{\mu}g/mg$, converted into the total lead concentration in groundwater ranging from $0.3{\mu}g/l$ to $4.7{\mu}g/l$. A lag time between recharge events and peak particle movement at the sampled wells was estimated to range from 19 to 22 days. Due to the particulate nature of lead in groundwater, only the wells connected with the high K zones detect lead, causing the spatial variation. In a given well, lead concentration varies at different sampling times due to the variation in the initial amounts of lead-carrying particles introduced into the groundwater system during recharge events, the lag in particle transport and the dispersion of lead-carrying particles along the advective flowpaths.

  • PDF

X-선 회절분석을 이용한 일라이트-스멕타이트 기본입자의 적층성에 관한 연구

  • 강일모;문희수
    • Proceedings of the Mineralogical Society of Korea Conference
    • /
    • 2003.05a
    • /
    • pp.10-10
    • /
    • 2003
  • 일라이트-스멕타이트 혼합층광물(I-S)은 열역학적으로 상호 대립적인 두 가지 모델로 이해되고 있다. 첫째, MacEwan 결정자 모델은 I-S를 5-20개의 스멕타이트와 일라이트 층으로 구성된 결정자로 해석한다. 이러한 모델은 분산과 재응집 과정을 기초로 하는 X-선 회절분석(XRD)에서 기인한 것으로 Reynolds의 XRD 모델과 동일하다. 둘째, 기본입자 모델은 I-S를 물리적으로 분리될 수 있는 최소 입자인 기본입자가 $c^{*-}$축 방향으로 응집된 응집체로 해석한다. 이러한 모델은 분산 과정을 기초로 하는 주사전자현미경(TEM) 관찰에서 기인한 모델이다. 강일모 등(2002)은 이 두 가지 모델을 비교함으로써 1< $N_{F}$<100/% $S_{XRD}$ ( $N_{F}$=평균 기본입자 층개수, %$S_{XRD}$=XRD 분석을 통하여 측정된 팽창성)을 도출하였다. 이 식은 기본입자모델과 Eberl & Srodon(1988)이 제시한 최대 팽창성(%$S_{MAX}$)을 동시에 해석할 수 있게 해준다. %$S_{MAX}$는 XRD 모델에서는 고려하지 않는 I-S 결정자 상$\cdot$하부에 존재하는 두 개의 0.5nm 규산염층을 하나의 스멕타이트 층으로 간주하여 얻어진 팽창성이다. Srodon et al.(1992)은 %$S_{MAX}$=100/ $N_{F}$을 제시하였으며, 강일모 등(2002)은 %$S_{MAX}$는 기하학적으로 기본입자가 무한적층을 하였을 때 관찰되는 %$S_{XRD}$와 동일함을 밝힌 바 있다. 만약, XRD 분석을 위한 시료 준비과정에서 I-S 결정자가 분산되었다가 재응집을 한다면, XRD에서 관찰되는 결과는 일차적으로 기본입자의 적층성에 영향을 받게 된다. 따라서, 기본 입자의 적층성은 XRD 분석을 이용하여 I-S 구조를 해석하는데 매우 중요한 요인이다. 본 연구는 기본입자의 적층성을 정량화하기 위해 %$S_{XRD}$=A/ $N_{F}$ (0$S_{MAX}$=100/ $N_{F}$로부터 얼마나 벗어나 있는가는 지시해 준다 금성산화산암복합체에서 산출되는 11개 I-S 시료와 14개의 Drits et al.(1998) 자료로부터 1nA=-0.14 $N_{F}$+4.7의 실험식을 도출할 수 있었으며, 기본입자의 적층성은 일차적으로 기본입자의 두께에 의해 영향을 받는 것으로 관찰되었다. Nadeau(1985)는 기본입자두께분포로부터 I-S 결정자의 팽창성을 측정하기 위하여 Ps=$\Sigma$p(N)/N을 제시하였다(Ps=스멕타이트 층 비율, N=기본 입자 층개수, p(N)=N의 확율). 그러나 위식은 실질적으로 %$S_{MAX}$를 제공해주기 때문에 %$S_{XRD}$를 유추하는데는 부적합하다. 본 연구는 이를 변형하여 Ps=$\Sigma$p(N)A(N)/N을 제시하였다(A(N)=N에 대한 A값). 위의 실험식을 사용하여 헝가리산 Zempleni 시료(15%$S_{XRD}$)의 기본입자분포로부터 %$S_{XRD}$를 계산한 결과, 16%$S_{XRD}$의 결과값을 얻을 수 있었다. 따라서, 본 연구에서 도출한 관계식들이 유효함을 확인할 수 있었다.계식들이 유효함을 확인할 수 있었다.

  • PDF

A Case Study on Autobiography Writing Program for Older Adults (노인자서전 쓰기 교육 프로그램 사례 연구)

  • Ji, Haeng Jung;Han, Jungran;Park, Seong Hie
    • 한국노년학
    • /
    • v.31 no.2
    • /
    • pp.223-241
    • /
    • 2011
  • The purpose of this research is to analyze the effect of an autobiography writing program as a method of education for older adults. The authors participated in the processes of all 10 sessions of the autobiography writing program as instructors or program assistants at a senior welfare center in Seoul, 2007. Data used for analysis were learners' narrations in class, written autobiographies by learners, and researchers' memo-notes. We analyzed their narrative flows using Haenninen(2000)'s theory of narrative flow and reconstructed their categories of activities as life histories. 'The innen narration' constructed by learners was a spiritual process in which they have imbued their life and living conditions with meaning. The learners reminisced their lives as 'unreflective narration' in the beginning of education. However, as the class went on, they gradually reminisced them as "reflective narration" and finally developed them as 'meta-reflective narration' through the intentional reflection. Likewise, we could interpret the transition of learners into meta-reflecter as the completion of ego-integration. As a conclusion, we found the effects of the autobiography writing education for older adults in educational, social, and psychological aspects.

COMPARISON OF MICROLEAKAGE WITH THREE DIFFERENT ADHESIVE SYSTEMS (수 종의 복합레진 접착 시스템에서의 미세 누출의 비교)

  • Seok, Choong-Ki;Nam, Dong-Woo;Nam, Soon-Hyeun;Kim, Young-Jin;Kim, Hyun-Jung
    • Journal of the korean academy of Pediatric Dentistry
    • /
    • v.31 no.4
    • /
    • pp.636-644
    • /
    • 2004
  • Recently, self-etching adhesive system have been developed and bonding procedures simplified into one or two steps, which are simultaneously applied to both enamel and dentin. These systems are easy to use and have the potential for good clinical success. The purpose of this study is to evaluate in vitro the microleakage on the cementum/dentin and enamel walls in composite resin restoration of Class V cavities, regarding the use of different adhesive systems. 30 human premolars were divided into 3 groups. A standardized Class V preparation was prepared on the buccal and lingual surface of each premolar. The preparation were made parallel to the cementoenamel junctions, with the gingival half of the preparation extending 1mm apical to the cementoenamel junction. After adhesive system was applied to teeth as manufacture's recommendation, hybrid resin composite was filled in bulk into the preparation and light polymerized according to manufacturer's recommendations. Specimen were stored in distilled water at $37^{\circ}C$ for 5 days and thermocycled 1000 times ($5^{\circ}C{\pm}2^{\circ}C\;and\;55^{\circ}C{\pm}2^{\circ}C)$, then immersed in a 2% methylene blue solution for 12 hours. After sectioning mesio distally through the restorations, the degree of dye penetration was scored under a stereomicroscope at ${\times}\;25$ magnification. The data were analyzed statistically using t-test and one-way ANOVA. The results were as follows: ${\cdot}$ There is no adhesive system which can prevent microleakage perfectly. ${\cdot}$ There is significant difference in microleakage between enamel margin and dentin margin (p<0.0001). ${\cdot}$ In enamel margin, self-etching primer systems did not show any significant difference comparing total-etching system. In denin margin, self-etching primer systems did not show any significant difference comparing one-bottle adhesive system used in combination with total-etching.

  • PDF

Characterization of Groundwater Quality and Recharge using Periodic Measurements of Hydrogeochemical Parameters and Environmental Tracers in Basaltic Aquifers of Jeju Island (수리지구화학적 인자와 환경 추적자의 주기적 관측에 의한 제주도 지하수의 수질과 함양 특성 분석)

  • Koh, Dong-Chan;Cheon, Su-Hyun;Park, Ki-Hwa
    • Journal of Soil and Groundwater Environment
    • /
    • v.12 no.4
    • /
    • pp.60-71
    • /
    • 2007
  • Groundwater from public wells was monitored during one year with two month interval for hydrogeochemical parameters and chlorofluorocarbons (CFCs) as environmental tracers in Jeju Island. Concentrations of major cations and $SiO_2$ show variation less than 10% whereas $NO_3$ and dissolved oxygen (DO) showed larger variation though DO variation did not change oxic or suboxic condition. $NO_3$ concentration has no consistent seasonal pattern with the largest variation of 35%. Groundwater ages determined by CFCs became temporarily younger by 5 years in October for groundwater with ages of 15 to 25 years, which can be attributed to infiltrating water in rainy season. Compared to air temperature, groundwater temperature has much smaller variation with no phase difference, which can be accounted for by a two-component model consisting of infiltrating water from surface and deeper groundwater with negligible temperature variation. The relatively small variation in groundwater age and temperature indicates that groundwater recharge through fast flow-paths is much smaller compared with basal groundwater in terms of aquifer storage.

Geochemical Characteristics of the Outer-Shelf Muddy Sediments in the East China Sea (동중국해 외대륙붕해역 니질퇴적물의 지화학적 특성)

  • Youn, Jeung-Su;Byun, Jong-Cheol;Kim, Yeo-Sang
    • Journal of the Korean earth science society
    • /
    • v.27 no.2
    • /
    • pp.198-208
    • /
    • 2006
  • To investigate the provenance of outer-shelf mud patch in the East China Sea, the geochemical compositions were analyzed and compared with those of Chinese rivers sediments. The mud sedimentary facies are distributed in the central region and sandy mud facies are also widely distributed around the study area. The major elements (Fe, Mg, K, Ti, and Mn) show strong positive correlation with Al, and trace elements also indicate the same characteristics; hence, clay minerals are likely to be the promising host for those elements. The high concentration of Fe, Ti, and Mn elements are found in the western middle part near the Changjiang estuary, indicating that it seems to result from the influence of the Changjiang River. Elemental ratios including Sc/Al, Ti/Nb, Th/Sc, Cr/Th, Nb/Co, and Th/U were thus used as provenance indicators to identify the sediment origins of the East China Sea. The discrimination diagrams clearly show that most of the sediment in the northern part are originated from the Huanghe River, while the muddy sediments in the western part near the Changjiang estuary might come from the Changjiang River, suggesting that the outer-shelf muddy sediments of East China Sea are originated from diverse sources.

A Development of Generalized Coupled Markov Chain Model for Stochastic Prediction on Two-Dimensional Space (수정 연쇄 말콥체인을 이용한 2차원 공간의 추계론적 예측기법의 개발)

  • Park Eun-Gyu
    • Journal of Soil and Groundwater Environment
    • /
    • v.10 no.5
    • /
    • pp.52-60
    • /
    • 2005
  • The conceptual model of under-sampled study area will include a great amount of uncertainty. In this study, we investigate the applicability of Markov chain model in a spatial domain as a tool for minimizing the uncertainty arose from the lack of data. A new formulation is developed to generalize the previous two-dimensional coupled Markov chain model, which has more versatility to fit any computational sequence. Furthermore, the computational algorithm is improved to utilize more conditioning information and reduce the artifacts, such as the artificial parcel inclination, caused by sequential computation. A generalized 20 coupled Markov chain (GCMC) is tested through applying a hypothetical soil map to evaluate the appropriateness as a substituting model for conventional geostatistical models. Comparing to sequential indicator model (SIS), the simulation results from GCMC shows lower entropy at the boundaries of indicators which is closer to real soil maps. For under-sampled indicators, however, GCMC under-estimates the presence of the indicators, which is a common aspect of all other geostatistical models. To improve this under-estimation, further study on data fusion (or assimilation) inclusion in the GCMC is required.

Characterizing Fluorescence Properties of Dissolved Organic Matter for Water Quality Management of Rivers and Lakes (하천 및 호소 수질관리를 위한 용존 자연유기물질 형광특성 분석)

  • Hur, Jin;Shin, Jae-Ki;Park, Sung-Won
    • Journal of Korean Society of Environmental Engineers
    • /
    • v.28 no.9
    • /
    • pp.940-948
    • /
    • 2006
  • Fluorescence measurements of dissolved organic matter(DOM) have the superior advantages over other analysis tools for applying to water quality management. They are simple and fast and require minimal pretreatment of samples. Fluorescence index($F_{450}/F_{500}$), synchronous spectra, and fluorescence excitation-emission matrices(EEM) of various DOM samples were investigated to discriminate autochthonous/allochthonous composition, protein-like fluorescence, fulvic-like fluorescence, humic-like fluorescence, terestrial humic-like fluorescence by comparing among the real DOM samples of different origins with the help of literature. The samples used included standard purified DOM, lake, river and wastewater treatment effluent. The relative distribution of various DOM composition was derived from the ratios of each fluorescence region. The results were very consistent with those expected from the sample properties. Allochthonous and terrestrial humic-like fluorescence were more prominent in the samples with abundant soil-derived DOM components. In addition, the protein-like fluorescence property was more pronounced in the samples where strong algal or microbial activities were expected. It was also shown that the ratio of protein-like/terrestrial humic-like fluorescence obtained from synchronous spectrum and fluorescence EEM could be used as an indicator for the evaluation of wastewater treatment on the downstream water quality of rivers and for the prediction of the degree of algal/microbial activities in lakes. It is expected that the results of this study will provide the basic information to develop the future water quality management techniques using DOM fluorescence measurements.

The Selective p-Distribution for Adaptive Refinement of L-Shaped Plates Subiected to Bending (휨을 받는 L-형 평판의 적응적 세분화를 위한 선택적 p-분배)

  • Woo, Kwang-Sung;Jo, Jun-Hyung;Lee, Seung-Joon
    • Journal of the Computational Structural Engineering Institute of Korea
    • /
    • v.20 no.5
    • /
    • pp.533-541
    • /
    • 2007
  • The Zienkiewicz-Zhu(Z/Z) error estimate is slightly modified for the hierarchical p-refinement, and is then applied to L-shaped plates subjected to bending to demonstrate its effectiveness. An adaptive procedure in finite element analysis is presented by p-refinement of meshes in conjunction with a posteriori error estimator that is based on the superconvergent patch recovery(SPR) technique. The modified Z/Z error estimate p-refinement is different from the conventional approach because the high order shape functions based on integrals of Legendre polynomials are used to interpolate displacements within an element, on the other hand, the same order of basis function based on Pascal's triangle tree is also used to interpolate recovered stresses. The least-square method is used to fit a polynomial to the stresses computed at the sampling points. The strategy of finding a nearly optimal distribution of polynomial degrees on a fixed finite element mesh is discussed such that a particular element has to be refined automatically to obtain an acceptable level of accuracy by increasing p-levels non-uniformly or selectively. It is noted that the error decreases rapidly with an increase in the number of degrees of freedom and the sequences of p-distributions obtained by the proposed error indicator closely follow the optimal trajectory.