• Journal of the Korean Statistical Society
• /
• v.29 no.1
• /
• pp.95-108
• /
• 2000

Consider the oblique factor model X=Af＋$\varepsilon$, with defining relation $\Sigma$=Λ$\Phi$Λ'＋Ψ. This paper is concerned with suggesting an optimal Bayes criterion for determining the number of factors in the model, i.e. dimension of the vector f. The use of marginal likelihood as a method for calculating posterior probability of each model with given dimension is developed under a generalized conjugate prior. Then based on an appropriate loss function, a Bayes rule is developed by use of the posterior probabilities. It is shown that the approach is straightforward to specify distributionally and to imploement computationally, with output readily adopted for constructing required cirterion.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.2
• /
• pp.263-283
• /
• 2024

Users' comments after online shopping are critical to product reputation and business improvement. These comments, sometimes known as e-commerce reviews, influence other customers' purchasing decisions. To confront large amounts of e-commerce reviews, automatic analysis based on machine learning and deep learning draws more and more attention. A core task therein is sentiment analysis. However, the e-commerce reviews exhibit the following characteristics: (1) inconsistency between comment content and the star rating; (2) a large number of unlabeled data, i.e., comments without a star rating, and (3) the data imbalance caused by the sparse negative comments. This paper employs Bidirectional Encoder Representation from Transformers (BERT), one of the best natural language processing models, as the base model. According to the above data characteristics, we propose the F_MixBERT framework, to more effectively use inconsistently low-quality and unlabeled data and resolve the problem of data imbalance. In the framework, the proposed MixBERT incorporates the MixMatch approach into BERT's high-dimensional vectors to train the unlabeled and low-quality data with generated pseudo labels. Meanwhile, data imbalance is resolved by Focal loss, which penalizes the contribution of large-scale data and easily-identifiable data to total loss. Comparative experiments demonstrate that the proposed framework outperforms BERT and MixBERT for sentiment analysis of e-commerce comments.

• Journal of Korean Ophthalmic Optics Society
• /
• v.9 no.1
• /
• pp.35-41
• /
• 2004

This study investigated the crystal growth, crystalline structure and the basic optical properties of $PbSnSe/BaF_2$ epilayers. The PbSnSe epilayer was grown on $BaF_2$(111) insulating substrates using a hot wall epitaxy (HWE) technique. It was found from the analysis of X-ray diffraction patterns that $PbSnSe/BaF_2$ epilayer was grown single crystal with a rock-salt structure oriented along [111] the growth direction. Using Rutherford back scattering, the atomic ratios of the PbSnSe was found to be proper stoichiometric. The best values for the full width at half maximum (FWHM) of the DCXRD was 162 arcsec for PbSnSe epilayer. The epilayer-thickness dependence of the FWHM of the DCXRD shows that the quality of the $PbSnSe/BaF_2$ is as expected. The dielectric function ${\varepsilon}(E)$ of a semiconductor is closely related to its electronic energy band structure and such relation can be drawn from features around the critical points(CPs) in the optical spectra. The real and imaginary parts(${\varepsilon}1$ and ${\varepsilon}2$) of the dielectric function ${\varepsilon}$ of PbSe were measured, and the observed spectra reveal distinct structures at energies of the E1, E2 and E3 CPs. These data are analyzed using a theoretical model known as the model dielectric function (MDF). The optical constants related to dielectric function such as the complex refractive index ($n^*=n+ik$), absorption coefficient (${\alpha}$) and normal-incidence reflectivity (R) are also presented for $PbSnSe/BaF_2$.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.14 no.4
• /
• pp.1561-1567
• /
• 2013

This paper is regarding the helicopter flight control law design for the handling quality performance. MIL-F-83300 and ADS-33E specification is used of the helicopter flight handling quality and to meet these requirements, ACAH type controller is required. This paper described the ACAH type controller design and performance evaluations. Helicopter dynamics first developed as nonlinear dynamics including rotor dynamics and then linear model was extracted from hovering to forward flight mode using trim condition. Control law used the model following to meet the handling qualities, the simple inverse model as feed forward gain, decoupling logic and phase model to decouple the axes, and linear model to calculate the coefficients. Handling quality evaluation used the matlab based Conduit tool and verified that Level 1 requirement is satisfied.

• Proceedings of the KSME Conference
• /
• 2008.11b
• /
• pp.2131-2136
• /
• 2008

The characteristics of heat transfer and pressure drop in an offset strip fin heat exchanger was studied with a steady-state three dimensional numerical model. Flow Reynolds number $Re_{dh}$ ranged from 10 to 3500 and Prandtl number Pr ranged from 0.7 to 50. The dimensionless performance factors, i.e. 'the pumping power factor F' and 'the heat transfer performance factor J' were analyzed and obtained a relationship between them. Finally, the prediction of F and J factors was generalized for different Prandtl numbers.

• Economic and Environmental Geology
• /
• v.39 no.2 s.177
• /
• pp.163-171
• /
• 2006

This study is the case of groundwater development based on the geometrical fracture model of target area established only through geological fracture mapping technique. A fracture mapping of $9km^2$, eastern Munsan, has been conducted to determine geological and hydrological factors for new water well placement in the Gyeonggi gneiss complex. Geophysical exploration was not applicable because of small restricted area and dense underground utilities at the site. Form line mapping on the basis of foliation orientation and rock type revealed a synform of NS fold axis bearing to the south. An EW geological cross-section passed through the site area shows a F2 synform as a double-wall ice cream spoon shape. Three regional faults of $N20^{\circ}E,\;N30^{\circ}W$, and NS have been dragged into the site to help understand extensional fault paths. The $N20^{\circ}E$ fault with dextral sense is geometrically interpreted as a western fault of two flexural conjugate type-P shear faults in the F2 synformal fold. The NE cross-section reveals that a possible groundwater belt in the western limb of super-posed fold area is formed as a trigonal prism within 100 m depth of the intersectional space between the $N20^{\circ}E$ fault plane and the weakly sheared plane of transposed foliation. Another possible fault for water resource strikes $N40^{\circ}E$. Recommended sites for new water well placement are along the $N20^{\circ}E\;and\;N40^{\circ}E$ faults. As a result of fracture mapping, 145 ton/day of water can be produced at one well along the $N20^{\circ}E$ fault line. Exploration of groundwater in the area is succeeded only using with geological fracture mapping and interpretation of geological cross-section, without any geophysical survey. Intersection of fault generated with the F2 synformal fold and foliation supply space of groundwater reserver.

• Structural Engineering and Mechanics
• /
• v.20 no.1
• /
• pp.45-67
• /
• 2005

This paper presents a new approach to predict the racking load-displacement response of plasterboard clad walls found in Australian light-framed residential structures under monotonic racking load. The method is based on a closed-form mathematical model, described herein as the 'Modularised' Closed-Form Mathematical model or MCFM model. The model considers the non-linear behaviour of the connections between the plasterboard cladding and frame. Furthermore, the model is flexible as it enables incorporation of different nailing patterns for the cladding. Another feature of this model is that the shape of stud deformation is not assumed to be a specific function, but it is computed based on the strain energy approach to take account of the actual load deformation characteristics of particular walls. Verification of the model against the results obtained from a detailed Finite Element (FE) model is also reported. Very good agreement between the closed form solution and that of the FE model was achieved.

• Proceedings of the Korean Geotechical Society Conference
• /
• 2008.03a
• /
• pp.1178-1187
• /
• 2008

Quality assurance for embankment compaction is one of very important procedures to guarantee high quality construction. However, only sand replacement method (KS F2312) and static plate load test (KS F2310) which are conventional and tiresome methods are used to evaluate degree of compaction at construction fields. Recently, new types of devices such as the geogauge and the light falling weight deflectometer (LFWD), the soil impact hammer (CASPFOL) and dynamic cone penetration test etc. which are able to substitute for the conventional methods are begun to use to evaluate soil stiffness. In this study, a laboratory model test was performed to evaluate correlations among test results obtained from the new devices and to assess the potential use of them. All test results have correlations with relative density and water content. Especially, the coefficients of correlation between $E_G$ from the geogauge and $K_{30'}$ from the soil impact hammer and between $E_G$ from the geogauge and $E_{LFWD}$ from LFWD are more than 0.7 but those between the results from DCP and others are less than those between $E_{G{\cdot}}$ and $K_{30'}$ and $E_G$ and $E_{LFWD}$.

• Journal of the Korean Geographical Society
• /
• v.39 no.6 s.105
• /
• pp.907-921
• /
• 2004

Lots of characteristics such as dwelling, neighborhood, and accessibility characteristics affect to the housevalue. Many researches have been done to identify values of each characteristic using hedonic technique. However, there is a limit to identify interaction of each characteristic and variation of each characteristic among the accessibility context. This paper has implemented the Expansion Method research paradigm to model the housevalue determination process in the city of Seoul. The findings of this paper have revealed the presence of contextual variations in the housevalue determination process. The initial model for housevalue reveals that as $F_1$ increases (i.e., larger the number of rooms/bathrooms, larger parking space) and/or $F_2$ increases (i.e., higher owner occupied housing units, higher apartment housing units) and/or $F_3$ increases, (i.e., higher the ratio of higher than college graduated households, 8 school zone, older housing units) the estimated housevalue increases. However, the above relationships drift across their respective contexts. The houses which have negative $F_1$ value, the housevalue does not fluctuate according to the distance to the city center or subcenters. However, the houses which have positive $F_1$ value, the closer to the subcenters or shorter to the river, the higher the estimated housevalues. On the other hand, in areas far from the subcenters, the estimated housevalues does not fluctuate much according to the corresponding $F_2$ level. In areas close to the subcenters, the estimated housevalues vary tremendously according to the $F_2$ value. In the residual analysis, it is revealed that large apartment which are located in Kangnam, IchongDong, MokDong are underestimated. This paper has contributed to our understanding of the housevalue determination process by providing an alternative conceptualization to the traditional approach.

• The Korean Journal of Applied Statistics
• /
• v.31 no.5
• /
• pp.667-675
• /
• 2018

When a financial time series consists of daily (closing) returns, traditional volatility models such as autoregressive conditional heteroskedasticity (ARCH) and generalized ARCH (GARCH) are useful to figure out daily volatilities. With high frequency returns in a day, one may adopt various multivariate GARCH techniques (MGARCH) (Tsay, Multivariate Time Series Analysis With R and Financial Application, John Wiley, 2014) to obtain intraday volatilities as long as the high frequency is moderate. When it comes to the ultra high frequency (UHF) case (e.g., one minute prices are available everyday), a new model needs to be developed to suit UHF time series in order to figure out continuous time intraday-volatilities. Aue et al. (Journal of Time Series Analysis, 38, 3-21; 2017) proposed functional GARCH (fGARCH) to analyze functional volatilities based on UHF data. This article introduces fGARCH to the readers and illustrates how to estimate fGARCH equations using UHF data of KOSPI and Hyundai motor company.