• Journal of KIISE
• /
• v.44 no.2
• /
• pp.179-185
• /
• 2017

Document summarization aims to generate a summary that is consistent and contains the highly related sentences in a document. In this study, we implemented for document summarization that extracts highly related sentences from a whole document by considering both similarities and entailment relations between sentences. Accordingly, we proposed a new algorithm, TextRank-NLI, which combines a Recurrent Neural Network based Natural Language Inference model and a Graph-based ranking algorithm used in single document extraction-based summarization task. In order to evaluate the performance of the new algorithm, we conducted experiments using the same datasets as used in TextRank algorithm. The results indicated that TextRank-NLI showed 2.3% improvement in performance, as compared to TextRank.

• Environmental and Resource Economics Review
• /
• v.21 no.3
• /
• pp.493-517
• /
• 2012

This paper analyzes a price discovery process for gasoline among branded and independent stations in Korea using a vector error correction model (VECM) and directed acyclic graphs (DAG). Two data sets for daily prices of medium level gasoline running from April 15, 2008 to May 31, 2009 and from January 1, 2011 to December 31, 2011 are used for empirical analysis. Empirical results show that S-OIL has an exogeneity and played a important role in the flow of price information in the market in the first period. In the second period, SK energy played a key role in price discovery process in the market. The price of NH-OIL stations do not cause the price of any other stations, which implies that the entrance of new branded stations with lower gasoline price to market has no influence on gasoline prices of retail markets.

• Journal of the Korea Society of Computer and Information
• /
• v.12 no.6
• /
• pp.105-113
• /
• 2007

A gene regulatory network is a network of genes representing how genes influence the activities of other genes. Nowadays from microarray experiments, a large number of measurements on the expression levels of genes are available. One of typical data is the so-called "steady-state model" data measuring the expression levels of other genes after knocking out a particular gene. This paper shows how to reverse engineer a parsimonious gene regulatory network, using these measurement data. Our model considers auto-regulation, which forms a cycle in a genetic network. We also model repressor and enhancer roles of genes. which are not considered in previous known methods.

• Journal of Biomedical Engineering Research
• /
• v.22 no.2
• /
• pp.133-138
• /
• 2001

동맥계에서 맥파의 전달속도는 혈관의 긴장도에 따라 결정된다. 동맥혈관의 긴장도는 혈압에 의존적이며 순간적인 혈압의 증가에 따른 혈관긴장도의 변화는 순환기계의 질병을 야기할 수 있다. 본 연구에서는 말초맥파를 이용한 동맥 긴장도 평가 방법의 유용성을 밝히기 위하여 말초맥파 도달시간을 체위에 따라 측정하고, 체위 변화에 따른 정수압 변화와 말초맥파도달시간의 관계를 관찰하였다. 말초맥파도달시간의 측정은 심전도의 QRS파 피크로부터 말초에서 나타나는 광전용적맥파의 최대기울기점까지 시간을 측정하였다. 생체 신호를 아날로그-디지털 변환하여 PC에서 매번 심장주기마다 맥파도달시간을 분석하였으며 결과는 스플라인 보간법을 사용하여 그래프로 결과를 제시하였다. 소프트웨어는 C＋＋을 사용하여 windows 환경에서 운용하였다. 앙아위에서 말초맥파도달시간은 233.38$\pm$1.84 ms이었으며 서있는 자세에서는 226.77$\pm$2.16 ms로 유의하게 감소하였다. 앙아위의 자세에서 손을 수직으로 들어올림에 따라 말초맥파도달시간은 각도에 비례하여 증가하는 특성을 보였다. 이러한 실험결과는 말초맥파도달시간이 동맥계의 일시적인 압력 변화에 매우 민감하게 반응하는 것을 제시하였으며, 고혈압 및 동맥경화가 있는 환자에서 예후를 판단하는 방법으로 유용하게 사용될 것이라 사료된다.

• Journal of the Korean Chemical Society
• /
• v.37 no.8
• /
• pp.730-734
• /
• 1993

The electrochemical reduction behavior or Biliverdin (BV) in N,N-dimethylformamide solvent was studied by DC polarography, cyclic voltammetry and the controlled potential coulometry. The reduced product was indentified by UV-Vis spectroscopy. In DC polarogram, two reduction waves of BV were founded. The half wave potentials of two reduction waves were -0.71 and -0.91 V vs. Ag/Ag$^+$ respectively. The current type of the 1st reduction wave was diffusion-controlled and the 2nd was diffusion current containing a little kinetic current. The 1st electrochemical reduction process was irreversible and BV reduced to Bilirubin.

• Journal of the Korean Chemical Society
• /
• v.37 no.10
• /
• pp.888-894
• /
• 1993

The electrochemical behaviors of copper-1-(2-pyridylazo)-2-naphthol(Cu-PAN) complex in acetonitrile (AN) solution have been investigated by the use of cyclic voltammetry, DC-polarography, controlled potential coulometry and UV-Vis spectroscopy. Cu-PAN complex in acetonitrile exhibit three reduction waves at -1.27 V, -1.64 V and -2.08 V vs. Ag/AgNO$_3$(AN). The numbers of electron involved in each reduction step was calculated with controlled potential coulometry, and reduction product was identified with UV-Vis spectrum. As the result, we proposed the reduction mechanism of the Cu-PAN complex in acetonitrile.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.17 no.5
• /
• pp.684-690
• /
• 2007

We describe our method to predict the direction of conditional probabilistic dependencies between clusters of random variables. Selected variables called 'gateway variables' are used to predict the conditional probabilistic dependency relations between clusters. The direction of conditional probabilistic dependencies between clusters are predicted by finding directed acyclic graph (DAG)-shaped dependency structure between the gateway variables. We show that our method shows meaningful prediction results in determining directions of conditional probabilistic dependencies between clusters.

• Journal of the Korea Computer Graphics Society
• /
• v.27 no.5
• /
• pp.73-79
• /
• 2021

Recently, in the field of computer animation, a method for generating motion using deep learning has been studied away from conventional finite-state machines or graph-based methods. The expressiveness of the network required for learning motions is more influenced by the diversity of motion contained in it than by the simple length of motion to be learned. This study aims to find an efficient network structure when the types of motions to be learned are diverse. In this paper, we train and compare three types of networks: basic fully-connected structure, mixture of experts structure that uses multiple fully-connected layers in parallel, recurrent neural network which is widely used to deal with seq2seq, and transformer structure used for sequence-type data processing in the natural language processing field.

• Journal of Wetlands Research
• /
• v.24 no.1
• /
• pp.59-67
• /
• 2022

The interest in LID facilities is increasing worldwide for recovery of natural water cycle system to destroy by urbanization. However, problems are raised when installation of LID because comprehensive analysis about design capacity adequacy of LID facilities was not conducted completely. In this research, removal efficiency and design volume adequacy of LID facilities were analyzed based on rainfall monitoring data in four LID facilities(Vegetated Swale, Vegetative Filter Strip, Bio-Retention and Permeable Pavement). As a result, group of LID facility designed on WQV was shown higher flow(37%) and pollutants(TSS, BOD, TN and TP) removal efficiencies(20 ~ 37%) than group of LID facility designed on WQF. SA/CA graph was drawn for evaluation of design volume adequacy based on rainfall monitoring data. In this SA/CA graph, coefficient of determination show over 0.5 in all parameter, especially, Flow and TP were show over 0.95. And, 'SA/CA & L/CA' graph considering difference of structure mechanism in LID facility suggested in this research was confirmed that improved coefficient of determination in flow, TSS and TP than SA/CA graph. According to this research results, feasibility of applying 'SA/CA & L/CA' graph for evaluation of design volume adequacy in LID facility, and it is necessary to follow up research for generalization and normalization.

• Journal of Aerospace System Engineering
• /
• v.13 no.6
• /
• pp.9-16
• /
• 2019

The experiment on the liquid jet in crossflow in supersonic BFS (backward-facing step) flow was conducted to investigate the mixing characteristics. The working fluids are nitrogen and water. The shadow graph technique was used to visualize the flow field. Images captured by the high-speed camera were applied to analyze the flow phenomena. The liquid jet was injected at the re-circulation zone created by the supersonic jet flow. Experimental conditions are defined based on the pressure of the nitrogen gas chamber and pressurized liquid tank. In respective cases, the penetration depth of liquid jet and location of the Mach disc were observed to be proportional to the momentum ratio of gas and liquid jets.