• Journal of the Korea Society of Computer and Information
• /
• v.27 no.12
• /
• pp.59-68
• /
• 2022

This paper presents a method for 1:1 verification by comparing the similarity between the given real product image and the drawing image. The proposed method combines two existing CNN-based deep learning models to construct a Siamese Network. After extracting the feature vector of the image through the FC (Fully Connected) Layer of each network and comparing the similarity, if the real product image and the drawing image (front view, left and right side view, top view, etc) are the same product, the similarity is set to 1 for learning and, if it is a different product, the similarity is set to 0. The test (inference) model is a deep learning model that queries the real product image and the drawing image in pairs to determine whether the pair is the same product or not. In the proposed model, through a comparison of the similarity between the real product image and the drawing image, if the similarity is greater than or equal to a threshold value (Threshold: 0.5), it is determined that the product is the same, and if it is less than or equal to, it is determined that the product is a different product. The proposed model showed an accuracy of about 71.8% for a query to a product (positive: positive) with the same drawing as the real product, and an accuracy of about 83.1% for a query to a different product (positive: negative). In the future, we plan to conduct a study to improve the matching accuracy between the real product image and the drawing image by combining the parameter optimization study with the proposed model and adding processes such as data purification.

• Journal of the Korea Convergence Society
• /
• v.13 no.3
• /
• pp.77-89
• /
• 2022

It has been proven through many previous researches that the pretrained language model with a large corpus helps improve performance in various natural language processing tasks. However, there is a limit to building a large-capacity corpus for training in a language environment where resources are scarce. Using the Cross-lingual Post-Training (XPT) method, we analyze the method's efficiency in Korean, which is a low resource language. XPT selectively reuses the English pretrained language model parameters, which is a high resource and uses an adaptation layer to learn the relationship between the two languages. This confirmed that only a small amount of the target language dataset in the relationship extraction shows better performance than the target pretrained language model. In addition, we analyze the characteristics of each model on the Korean language model and the Korean multilingual model disclosed by domestic and foreign researchers and companies.

• Korean Journal of Optics and Photonics
• /
• v.33 no.1
• /
• pp.28-34
• /
• 2022

We study through computational simulation the focal performance of an infrared (IR) Fresnel lens, composed of a multilayer-graphene zone plate formed under a graphene electrode. Here the Fermi level E_F of the patterned multilayer graphene is adjusted by the overlying graphene electrode. The Fresnel lens effect, with respect to the reflectance contrast between the graphene electrode and the 8-layer graphene zone plate placed on a glass substrate, has been analyzed over a broad wavelength range from 4 to 30 ㎛. As the optimal wavelength of 8 ㎛ (considering the reflectance and the reflectance-contrast ratio) is incident upon the Fresnel lens with a focal length of 240 ㎛, the focal intensity is enhanced by a factor of 4.3 as the E_F of multilayer graphene increases from 0.4 eV to 1.6 eV, and is improved by a factor of 5.8 as the number of graphene layers increases from two to eight. As a result, an all-graphene-based IR Fresnel zone-plate lens, exhibiting multifocal function (240 ㎛ and 360 ㎛) according to the selected E_F, is proposed as an ultrathin lens platform.

• Knowledge Management Research
• /
• v.23 no.2
• /
• pp.277-299
• /
• 2022

The massive card transaction data generated in the tourism industry has become an important resource that implies tourist consumption behaviors and patterns. Based on the transaction data, developing a smart service system becomes one of major goals in both tourism businesses and knowledge management system developer communities. However, the lack of rating scores, which is the basis of traditional recommendation techniques, makes it hard for system designers to evaluate a learning process. In addition, other auxiliary factors such as temporal, spatial, and demographic information are needed to increase the performance of a recommendation system; but, gathering those are not easy in the card transaction context. In this paper, we introduce CTDDTR, a novel approach using card transaction data to recommend tourism services. It consists of two main components: i) Temporal preference Embedding (TE) represents tourist groups and services into vectors through Doc2Vec. And ii) Deep tourism Recommendation (DR) integrates the vectors and the auxiliary factors from a tourism RDF (resource description framework) through MLP (multi-layer perceptron) to provide services to tourist groups. In addition, we adopt RFM analysis from the field of knowledge management to generate explicit feedback (i.e., rating scores) used in the DR part. To evaluate CTDDTR, the card transactions data that happened over eight years on Jeju island is used. Experimental results demonstrate that the proposed method is more positive in effectiveness and efficacies.

• Journal of Conservation Science
• /
• v.38 no.3
• /
• pp.243-253
• /
• 2022

Using non-destructive analytical methods, we identified the material characteristics of two gold fabric artifacts excavated from the Cheongsong Sim clan (Bugeum Wonsam, Jikgeum Chima), including the artifact condition, fiber type, surface contamination, and metallic threads. We found that the artifacts were buried and had turned brown; thus, we were unable to determine their original color. The fiber type was determined to be silk from cocoons, based on scanning electron microscopy, Fourier transform infrared (FT-IR) analyses of Amide I, II, III, and IV peaks, and color reactions Further, the FT-IR and X-ray fluorescence (XRF) analyses identified the white and black stains as natural resin hydrolyzed substances, such as lipids and proteins, that occurred as microbial decomposition due to body decay. Finally, the XRF analyses identified the thin gold layer of the metallic yarn as gold (Au). According to the FT-IR data and the color reaction to the metallic yarn medium, the adhesive component of the medium was a product of-Amides I, II, III, and 3000 cm^-1 within Amides A and B (an animal type), respectively. Thus, the medium was identified as Hanji (Korean paper), which is made from domestically produced Broussonetia kazinoki fibers.

• Explosives and Blasting
• /
• v.40 no.4
• /
• pp.15-26
• /
• 2022

With the recent increase in old and dangerous buildings, the demand for technology in the field of structure demolition is rapidly increasing. In particular, in the case of structures with severe deformation of damage, there is a risk of deterioration in stability and disaster due to changes in the load distribution characteristics in the structure, so rapid structure demolition technology that can be efficiently dismantled in a short period of time is drawing attention. However, structural deformation such as unauthorized extension or illegal remodeling occurs frequently in many old structures, which is not reflected in structural information such as building drawings, and acts as an obstacle in the demolition design process. In this study, as an effective way to overcome the discrepancy between the structural information of old structures and the actual structure, access to actual structures through 3D modeling was considered. 3D point cloud data inside and outside the building were obtained through LiDAR and drone photography for buildings scheduled to be blasting demolition, and precision matching between the two spatial data groups was performed using an open-source based spatial information construction system. The 3D structure model was completed by importing point cloud data matched with 3D modeling software to create structural drawings for each layer and forming each member along the structure slab, pillar, beam, and ceiling boundary. In addition, the modeling technique proposed in this study was verified by comparing it with the actual measurement value for selected structure member.

• Journal of Wetlands Research
• /
• v.25 no.2
• /
• pp.121-131
• /
• 2023

This study examines the possible problem in the rainfall-runoff analysis process using the VIC (Variable Infiltration Capacity) model caused by using the effective soil depth instead of the soil depth. The parameters of the model are determined as follows. First, parameters that can be determined using available numerical information are fixed. For parameters related to direct runoff and base runoff, the recommended values of the VIC model are applied. In the case of soil depth, four cases are considered: (1) the effective soil depth is applied as the soil depth, (2) 1.5 times of the effective soil depth is applied as the soil depth by reflecting the vertical structure of the soil layer, (3) 1.25 times of the effective soil depth, and (4) 2.0 times of the effective soil depth as alternative soil depths. This study simulates the rainfall-runoff for the period from 1983 to 2020 targeting the Chungju Dam and Soyang River Dam basins of the Han River system. As a result of the study, it is confirmed that when the effective soil depth is applied instead of the soil depth, direct runoff and baseflow have opposite effects, and direct runoff increases by more than 3% while base runoff decreases by the same scale. In addition, the most influential factor in the estimation of the effective soil depth in the Chungju Dam and Soyanggang Dam basins is found to be the proportion of rock outcrop area. The difference between the direct runoff ratio and the base runoff ratio in the two basins is conformed significantly different due to the influence of the rock outcrop area.

• Geomechanics and Engineering
• /
• v.35 no.1
• /
• pp.29-39
• /
• 2023

As the intensity of urban underground space development increases, more and more tunnels are planned and constructed, and sometimes it is inevitable to encounter situations where tunnels have to underpass the river embankments. Most previous studies involved tunnels passing river embankments perpendicularly or with large intersection angle. In this study, a project case where two EPB shield tunnels with 8.82 m diameter run parallelly underneath a river embankment was reported. The parallel length is 380 m and tunnel were mainly buried in the moderate / slightly weathered clastic rock layer. The field monitoring result was presented and discussed. Three-dimensional back-analysis were then carried out to gain a better understanding the interaction mechanisms between shield tunnel and embankment and further to predict the ultimate settlement of embankment due to twin-tunnel excavation. Parametrical studies considering effect of tunnel face pressure, tail grouting pressure and volume loss were also conducted. The measured embankment settlement after the single tunnel excavation was 4.53 mm ~ 7.43 mm. Neither new crack on the pavement or cavity under the roadbed was observed. It is found that the more degree of weathering of the rock around the tunnel, the greater the embankment settlement and wider the settlement trough. Besides, the latter tunnel excavation might cause larger deformation than the former tunnel excavation if the mobilized plastic zone overlapped. With given geometry and stratigraphic condition in this study, the safety or serviceability of the river embankment would hardly be affected since the ultimate settlement of the embankment after the twin-tunnel excavation is within the allowable limit. Reasonable tunnel face pressure and tail grouting pressure can to some extent suppress the settlement of the embankment. The recommended tunnel face pressure and tail grouting pressure are 300 kPa and 550 kPa in this study, respectively. However, the volume loss plays the crucial role in the tunnel-embankment interaction. Controlling and compensating the tunneling induced volume loss is the most effective measure for river embankment protection. Additionally, reinforcing the embankment with cement mixing pile in advance is an alternative option in case the predicted settlement exceeds allowable limit.

• Korean Journal of Environmental Agriculture
• /
• v.42 no.3
• /
• pp.203-210
• /
• 2023

Ferimzone Z is a fungicide for effectively controlling rice blast. Under light irradiation conditions, it undergoes a rapid conversion to its E-stereoisomer. Given the importance of isomers in risk assessments of residues in crops, an analytical method was developed for individual isomer quantification. A comparative analysis performed using two columns in HPLC-MS/MS demonstrated that the isomers were successfully separated using the Cadenza column. For the brown rice sample preparation, 5 g of the homogenized sample was saturated with 7 mL of water. The sample was then extracted with a 10 mL mixed solvent of acetonitrile and ethyl acetate (1:1, v/v) that contained 0.1% formic acid, and it was subsequently partitioned with magnesium sulfate and sodium chloride. The upper layer was purified using dSPE containing C₁₈ and PSA sorbents. The established method was subjected to method validation, and it showed recovery rates of 90.6-98.8% (RSD ≤ 3.9%) at concentrations of 0.01, 0.1, 2 mg/kg, with a soft matrix effect (%ME) ranging from -3.1% to +6.5%. This method can be employed in monitoring studies of brown rice to determine the conversion ratio from the Z isomers to the E isomers.

• Journal of the Korean Geotechnical Society
• /
• v.39 no.9
• /
• pp.13-24
• /
• 2023

To investigate the stability of temporary retaining walls during excavation, it is essential to develop reverse analysis technologies capable of precisely evaluating the properties of the ground and a learning model that can assess stability by analyzing real-time data. In this study, we targeted excavation sites where the C.I.P method was applied. We developed a Deep Neural Network (DNN) model capable of evaluating the stability of the retaining wall, and estimated the physical properties of the ground being excavated using a Differential Evolution Algorithm. We performed reverse analysis on a model composed of a two-layer ground for the applicability analysis of the Differential Evolution Algorithm. The results from this analysis allowed us to predict the properties of the ground, such as the elastic modulus, cohesion, and internal friction angle, with an accuracy of 97%. We analyzed 30,000 cases to construct the training data for the DNN model. We proposed stability evaluation grades for each assessment factor, including anchor axial force, uneven subsidence, wall displacement, and structural stability of the wall, and trained the data based on these factors. The application analysis of the trained DNN model showed that the model could predict the stability of the retaining wall with an average accuracy of over 94%, considering factors such as the axial force of the anchor, uneven subsidence, displacement of the wall, and structural stability of the wall.