• The Korean Journal of Applied Statistics
• /
• v.28 no.5
• /
• pp.933-949
• /
• 2015

Audience rating competition in the domestic drama market has increased recently due to the introduction of commercial broadcasting and diversification of channels. There is now a need for thorough studies and analysis on audience rating. Especially, a drama rating is an important measure to estimate advertisement costs for producers and advertisers. In this paper, we study the drama rating prediction models using various data mining techniques such as linear regression, LASSO regression, random forest, and gradient boosting. The analysis results show that initial drama ratings are affected by structural elements such as broadcasting station and broadcasting time. Average drama ratings are also influenced by earlier public opinion such as the number of internet searches about the drama.

• Applied Chemistry for Engineering
• /
• v.16 no.6
• /
• pp.853-856
• /
• 2005

Inductively coupled plasma reactive ion etching of magnetic tunnel junction (MTJ) stack, which is one of the key elements in magnetic random access memory, was studied. The MTJ stacks were patterned in nanometer size by electron(e)-beam lithography, and TiN thin films were employed as a hard mask. The etch process of TiN hard mask was examined using Ar, $Cl_2/Ar$, and $SF_6/Ar$. The TiN hard mask patterned by e-beam lithography was first etched and then the etching of MTJ stack was performed. The MTJ stacks were etched using Ar, $Cl_2/Ar$, and $BCl_3/Ar$ gases by varying gas concentration and pressure.

• Geomechanics and Engineering
• /
• v.17 no.1
• /
• pp.31-45
• /
• 2019

In this study the spatial variability of soils is substantiated physically and numerically by using random field theory. Heterogeneous samples are fabricated by combining nine homogeneous soil clusters that are assumed to be elements of an adopted random field. Homogeneous soils are prepared by mixing different percentages of kaolin and bentonite at water contents equivalent to their respective liquid limits. Comprehensive characteristic laboratory tests were carried out before embarking on direct shear experiments to deduce the basic correlations and properties of nine homogeneous soil clusters that serve to reconstitute the heterogeneous samples. The tests consist of Atterberg limits, and Oedometric and unconfined compression tests. The undrained shear strength of nine soil clusters were measured by the unconfined compression test data, and then correlations were made between the water content and the strength and stiffness of soil samples with different consistency limits. The direct shear strength of heterogeneous samples of different stochastic properties was then evaluated by physical and numerical modelling using FISH code programming in finite difference software of $FLAC^{3D}$. The results of the experimental and stochastic numerical analyses were then compared. The deviation of numerical simulations from direct shear load-displacement profiles taken from different sources were discussed, potential sources of error was introduced and elaborated. This study was primarily to explain the mathematical and physical procedures of sample preparation in stochastic soil mechanics. It can be extended to different problems and applications in geotechnical engineering discipline to take in to account the variability of strength and deformation parameters.

• Ocean Systems Engineering
• /
• v.11 no.1
• /
• pp.17-42
• /
• 2021

Articulated loading platforms (ALPs) belongs to a class of offshore structures known as compliant. ALP motions have time periods falling in the wind excitation frequency range due to their compliant behaviour. This paper deals with the dynamic behavior of a double hinged ALP subjected to low-frequency wind forces with random waves. Nonlinear effects due to variable submergence, fluctuating buoyancy, variable added mass, and hydrodynamic forces are considered in the analysis. The random sea state is characterized by the Pierson-Moskowitz (P-M) spectrum. The wave forces on the submerged elements of the platform's shaft are calculated using Morison's Equation with Airy's linear wave theory ignoring diffraction effects. The fluctuating wind load has been estimated using Ochi and Shin wind velocity spectrum for offshore structures. The nonlinear dynamic equation of motion is solved in the time domain by the Wilson-θ method. The wind-structure interactions, along with the effect of various other parameters on the platform response, are investigated. The effect of offset of aerodynamic center (A.C.) with the center of gravity (C.G.) of platform superstructure has also been investigated. The outcome of the analyses indicates that low-frequency wind forces affect the response of ALP to a large extent, which otherwise is not enhanced in the presence of only waves. The mean wind modifies the mean position of the platform surge response to the positive side, causing an offset. Various power spectral densities (PSDs) under high and moderate sea states show that apart from the significant peak occurring at the two natural frequencies, other prominent peaks also appear at very low frequencies showing the influence of wind on the response.

• Nuclear Engineering and Technology
• /
• v.54 no.1
• /
• pp.49-60
• /
• 2022

Motivated by the high-resolution properties of high-order Weighted Essentially Non-Oscillatory (WENO) and flux limiter (FL) for steep-gradient problems and the robust convergence of Jacobian-free Newton-Krylov (JFNK) methods for nonlinear systems, the preconditioned JFNK fully implicit high-order WENO and FL schemes are proposed to solve the transient two-phase two-fluid models. Specially, the second-order fully-implicit BDF2 is used for the temporal operator and then the third-order WENO schemes and various flux limiters can be adopted to discrete the spatial operator. For the sake of the generalization of the finite-difference-based preconditioning acceleration methods and the excellent convergence to solve the complicated and various operational conditions, the random vector instead of the initial condition is skillfully chosen as the solving variables to obtain better sparsity pattern or more positions of non-zero elements in this paper. Finally, the WENO_JFNK and FL_JFNK codes are developed and then the two-phase steep-gradient problem, phase appearance/disappearance problem, U-tube problem and linear advection problem are tested to analyze the convergence, computational cost and efficiency in detailed. Numerical results show that WENO_JFNK and FL_JFNK can significantly reduce numerical diffusion and obtain better solutions than traditional methods. WENO_JFNK gives more stable and accurate solutions than FL_JFNK for the test problems and the proposed finite-difference-based preconditioning acceleration methods based on the random vector can significantly improve the convergence speed and efficiency.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.7
• /
• pp.1759-1772
• /
• 2023

Chinese named entity recognition (NER) is a challenging work that seeks to find, recognize and classify various types of information elements in unstructured text. Due to the Chinese text has no natural boundary like the spaces in the English text, Chinese named entity identification is much more difficult. At present, most deep learning based NER models are developed using a bidirectional long short-term memory network (BiLSTM), yet the performance still has some space to improve. To further improve their performance in Chinese NER tasks, we propose a new NER model, IDCNN-BiLSTM-Highway, which is a combination of the BiLSTM, the iterated dilated convolutional neural network (IDCNN) and the highway network. In our model, IDCNN is used to achieve multiscale context aggregation from a long sequence of words. Highway network is used to effectively connect different layers of networks, allowing information to pass through network layers smoothly without attenuation. Finally, the global optimum tag result is obtained by introducing conditional random field (CRF). The experimental results show that compared with other popular deep learning-based NER models, our model shows superior performance on two Chinese NER data sets: Resume and Yidu-S4k, The F1-scores are 94.98 and 77.59, respectively.

• The Journal of Korean Academy of Prosthodontics
• /
• v.49 no.3
• /
• pp.214-221
• /
• 2011

Purpose: To date most of finite element analysis assumed the presence of 100% contact between bone and implant, which is inconsistent with clinical reality. In human retrieval study bone-implant contact (BIC) ratio ranged from 20 to 80%. The objective of this study was to explore the influence of bone-implant contact pattern on bone of the interface using nonlinear 3-dimensional finite element analysis. Materials and methods: A computer tomography-based finite element models with two types of implant (Mark III Br${\aa}$nemark$^{(R)}$, Inplant$^{(R)}$) which placed in the maxillary 2nd premolar area were constructed. Two different degrees of bone-implant contact ratio (40, 70%) each implant design were simulated. 5 finite element models were constructed each bone-implant contact ratio and implant design, and sum of models was 40. The position of bone-implant contact was determined according to random shuffle method. Elements of bone-implant contact in group W (wholly randomized osseointegration) was randomly selected in terms of total implant length including cortical and cancellous bone, while ones in group S (segmentally randomized osseointegration) was randomly selected each 0.75 mm vertically and horizontally. Results: Maximum von Mises strain between group W and group S was not significantly different regardless of bone-implant contact ratio and implant design (P=.939). Peak von Mises strain of 40% BIC was significantly lower than one of 70% BIC (P=.007). There was no significant difference between Mark III Br${\aa}$nemark$^{(R)}$ and Inplant$^{(R)}$ in 40% BIC, while average of peak von Mises strain for Inplant$^{(R)}$ was significantly lower ($4886{\pm}1034\;{\mu}m/m$) compared with MK III Br${\aa}$nemark$^{(R)}$ ($7134{\pm}1232\;{\mu}m/m$) in BIC 70% (P<.0001). Conclusion: Assuming bone-implant contact in finite element method, whether the contact elements in bone were wholly randomly or segmentally randomly selected using random shuffle method, both methods could be effective to be no significant difference regardless of sample size.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.42 no.6 s.336
• /
• pp.31-38
• /
• 2005

In this paper, we proposed an optical security system based on a gray-image exclusive-OR encryption using multi-phase separation and phase-wrapping method. For encryption, a gray image is sliced into binary images, which have the same pixel value, and these images are encrypted by modified XOR rules with binary random images. The XORed images and the binary images respectively combined and converted into full phase images, called an encrypted image and a key image. For decryption, when the encrypted image and key image are used as inputs on optical elements, Practically due to limited controllability of phase range in optical elements, the original gray image cannot be efficiently reconstructed by these optical elements. Therefore, by decreasing the phase ranges of the encrypted image and key image using a phase-wrapping method and separating these images into low-level phase images using multi-phase separation, the gray image can be reconstructed by optical elements which have limited control range. The decrytion process is simply implemented by interfering a multiplication result of encrypted image and key image with reference light. The validity of proposed scheme is verified and the effects, which are caused by phase limitation in decryption process, is analyzed by using computer simulations.

• Journal of The Korean Association For Science Education
• /
• v.40 no.3
• /
• pp.321-336
• /
• 2020

We explore automated scoring models of scientific argumentation. We consider how a new analytical approach using a machine learning technique may enhance the understanding of spoken argumentation in the classroom. We sampled 2,605 utterances that occurred during a high school student's science class on molecular structure and classified the utterances into five argumentative elements. Next, we performed Text Preprocessing for the classified utterances. As machine learning techniques, we applied support vector machines, decision tree, random forest, and artificial neural network. For enhancing the identification of rebuttal elements, we used a heuristic feature-engineering method that applies experts' classification of morphemes of scientific argumentation.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.21 no.5
• /
• pp.697-703
• /
• 2018

In this paper, we analyzed the operation of the ignition circuit for electronic arm and fire device(EAFD), and investigated the sensitive elements of the circuit system. For reliability analysis, the EAFD ignition circuit was modeled using the PSpice simulation tool, and the output results of the circuit were examined by changing the tolerance of each circuit element. Monte Carlo simulation was used by maintaining the values of the observed sensitive elements at ${\pm}10%$ of the original values and adjusting the values of the other components according to a random distribution. The histogram results of the output peak currents and pulse widths were represented by Weibull and Burr type XII function fittings in three cases(element values are +10 %, 0 %, -10 % of original). For the output peak currents, mean values were 1.0028, 1.0034, and 1.0050, where the variance values were calculated as 0.0398, 0.0396, and 0.0290 using the Weibull function fitting, respectively. For pulse widths, the mean values of 0.9475, 0.9907, and 1.0293 with the variance values of 0.0260, 0.0251, and 0.0238 were obtained using the Burr Type XII function fittings.