• Journal of the Korea Convergence Society
• /
• v.12 no.12
• /
• pp.93-106
• /
• 2021

Blenderbot 2.0 is a dialogue model representing open domain chatbots by reflecting real-time information and remembering user information for a long time through an internet search module and multi-session. Nevertheless, the model still has many improvements. Therefore, this paper analyzes the limitations and errors of BlenderBot 2.0 from three perspectives: model, data, and dialogue. From the data point of view, we point out errors that the guidelines provided to workers during the crowdsourcing process were not clear, and the process of refining hate speech in the collected data and verifying the accuracy of internet-based information was lacking. Finally, from the viewpoint of dialogue, nine types of problems found during conversation and their causes are thoroughly analyzed. Furthermore, practical improvement methods are proposed for each point of view, and we discuss several potential future research directions.

• Journal of Broadcast Engineering
• /
• v.27 no.2
• /
• pp.228-239
• /
• 2022

In this paper, we propose an unmanned vehicle scratch detection deep learning model for car sharing services. Conventional scratch detection models consist of two steps: 1) a deep learning module for scratch detection of images before and after rental, 2) a manual matching process for finding newly generated scratches. In order to build a fully automatic scratch detection model, we propose a one-step unmanned scratch detection deep learning model. The proposed model is implemented by applying transfer learning and fine-tuning to the deep learning model that detects changes in satellite images. In the proposed car sharing service, specular reflection greatly affects the scratch detection performance since the brightness of the gloss-treated automobile surface is anisotropic and a non-expert user takes a picture with a general camera. In order to reduce detection errors caused by specular reflected light, we propose a preprocessing process for removing specular reflection components. For data taken by mobile phone cameras, the proposed system can provide high matching performance subjectively and objectively. The scores for change detection metrics such as precision, recall, F1, and kappa are 67.90%, 74.56%, 71.08%, and 70.18%, respectively.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.5
• /
• pp.162-170
• /
• 2021

The Smart City Living Lab is becoming important as a local innovation platform to develop urban solutions. In January 2018, the 4th industrial innovation committee, which was a direct subordinate from the president, empathized citizens' participation and their roles within the Smart City [Urban Innovation and Future Growth Engine-Creating Smart City Strategy]. This was the starting point of the living lab. The central government and local governments have been promoting various types of living labs to encourage citizens to participate. On the other hand, due to the lack of systematic concepts and theories for practicing and structuring living labs, the practice is not performed well. This study aimed to develop systematic approaches and implementation methods of the public-led Smart City Living Lab. The Full-cycle Smart City living Lab model was designed by integrating smart city living lab work processes, as suggested in the standards of the national land plan, double design diamond framework, which is a type of innovative design methodology, and design thinking process. The entire cycle Smart City living lab model requires four components to practice the living lab, such as framework, module, process, and methodologies. In the future, this model is expected to be incorporated in the Smart City Living Lab.

• Resources Recycling
• /
• v.30 no.2
• /
• pp.39-45
• /
• 2021

Spent photovoltaic module is one of the important resource of silver, while related research concerning silver recovery remains limited. In our previous research, HNO3 was utilized to dissolve Ag(I) and Al(III) from the spent silicon solar cells. In order to recover Ag(I) from the leachate of a silicon solar cell, the present study made use of a nitrate solution containing Ag(I) and Al(III), which was subjected to a solvent extraction process with 5,8-diethyl-7-hydroxydodecan-6-oxime (LIX63). Ag(I) was selectively extracted with LIX63 over Al(III) from the nitrate leach solution. Subsequently, quantitative stripping of Ag(I) from the loaded LIX63 was performed by using 20% ammonia water. The McCabe-Thiele plots for the extraction and stripping isotherms of Ag(I) were also constructed. Extraction and stripping simulation tests confirmed an Ag(I) extraction and stripping efficiency of >99.99% and 98.9%, respectively with high purity Ag (99.998%) and Al (99.99%) solution. A process flow sheet for Ag(I) recovery from the nitrate leach solution was proposed.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.1
• /
• pp.111-118
• /
• 2023

The purpose of this study is to study jewelry designs presented to general consumers who seek new products and diversity. We would like to present a modular jewelry design with a structure and combination method that is distinct from jewelry in a multimodal replacement method that allows various product modules sold in the past to be worn interchangeably. Problems are likely to occur when a number of existing rather small parts are manufactured in a complex combination method, and difficulties may follow when consumers replace decorative parts and lose them in the process of assembling small fixture parts. Therefore, in order to reduce these problems, we try to make it different from jewelry products made with a simple and simple design so that it can be easily replaced and worn without the need for other coupling parts, and produced using the latest 3D printer (Rapid Prototyping). In this study, based on the experience and know-how gained while engaging in field work, it was possible to make a real object and focused on minimizing problems during the production process, and through this, time and economic loss can be reduced. The purpose of the study is to produce improved jewelry products by expressing more sophisticated and differentiated shapes by using 3D programs (CAD).

• Membrane Journal
• /
• v.33 no.6
• /
• pp.352-361
• /
• 2023

Steam methane reforming is currently the most widely used technology for producing hydrogen, a clean fuel. Hydrogen produced by steam methane reforming contains impurities such as carbon monoxide, and it is essential to undergo an appropriate post-purification step for commercial usage, such as fuel cells. Recently, membrane separation technology has been gaining great attention as an effective purification method; in this study, we evaluated the feasibility of using commercial polysulfone membranes for biogas upgrading to separate and recover hydrogen from a hydrogen/carbon monoxide gas mixture. Initially, we examined the physicochemical properties of the commercial membrane used. We then conducted performance evaluations of the commercial membrane module under various conditions using mixed gas, considering factors such as stage-cut and operating pressure. Finally, based on the evaluation results, we carried out simulations for process design. The maximum H₂ permeability and H₂/CO separation factor for the commercial membrane process were recorded at 361 GPU and 20.6, respectively. Additionally, the CO removal efficiency reached up to 94%, and the produced hydrogen concentration achieved a maximum of 99.1%.

• Journal of the Microelectronics and Packaging Society
• /
• v.30 no.4
• /
• pp.61-68
• /
• 2023

In this study, Sn-3.5Ag-15.0Fe composite solder was manufactured and applied to TLP bonding to change the entire joint into a Sn-Fe IMC(intermetallic compound), thereby applying it as a high-temperature solder. The FeSn₂ IMC formed during the bonding process has a high melting point of 513℃, so it can be stably applied to power modules for power semiconductors where the temperature rises up to 280℃ during use. As a result of applying ENIG surface treatment to both the chip and substrate, a multi-layer IMC structure of Ni₃Sn₄/FeSn₂/Ni₃Sn₄ was formed at the joint. During the shear test, the fracture path showed that cracks developed at the Ni₃Sn₄/FeSn₂ interface and then propagated into FeSn₂. After 2hours of the TLP joining process, a shear strength of over 30 MPa was obtained, and in particular, there was no decrease in strength at all even in a shear test at 200℃. The results of this study can be expected to lead to materials and processes that can be applied to power modules for electric vehicles, which are being actively researched recently.

• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.1-25
• /
• 2020

In this paper, we suggest an application system architecture which provides accurate, fast and efficient automatic gasometer reading function. The system captures gasometer image using mobile device camera, transmits the image to a cloud server on top of private LTE network, and analyzes the image to extract character information of device ID and gas usage amount by selective optical character recognition based on deep learning technology. In general, there are many types of character in an image and optical character recognition technology extracts all character information in an image. But some applications need to ignore non-of-interest types of character and only have to focus on some specific types of characters. For an example of the application, automatic gasometer reading system only need to extract device ID and gas usage amount character information from gasometer images to send bill to users. Non-of-interest character strings, such as device type, manufacturer, manufacturing date, specification and etc., are not valuable information to the application. Thus, the application have to analyze point of interest region and specific types of characters to extract valuable information only. We adopted CNN (Convolutional Neural Network) based object detection and CRNN (Convolutional Recurrent Neural Network) technology for selective optical character recognition which only analyze point of interest region for selective character information extraction. We build up 3 neural networks for the application system. The first is a convolutional neural network which detects point of interest region of gas usage amount and device ID information character strings, the second is another convolutional neural network which transforms spatial information of point of interest region to spatial sequential feature vectors, and the third is bi-directional long short term memory network which converts spatial sequential information to character strings using time-series analysis mapping from feature vectors to character strings. In this research, point of interest character strings are device ID and gas usage amount. Device ID consists of 12 arabic character strings and gas usage amount consists of 4 ~ 5 arabic character strings. All system components are implemented in Amazon Web Service Cloud with Intel Zeon E5-2686 v4 CPU and NVidia TESLA V100 GPU. The system architecture adopts master-lave processing structure for efficient and fast parallel processing coping with about 700,000 requests per day. Mobile device captures gasometer image and transmits to master process in AWS cloud. Master process runs on Intel Zeon CPU and pushes reading request from mobile device to an input queue with FIFO (First In First Out) structure. Slave process consists of 3 types of deep neural networks which conduct character recognition process and runs on NVidia GPU module. Slave process is always polling the input queue to get recognition request. If there are some requests from master process in the input queue, slave process converts the image in the input queue to device ID character string, gas usage amount character string and position information of the strings, returns the information to output queue, and switch to idle mode to poll the input queue. Master process gets final information form the output queue and delivers the information to the mobile device. We used total 27,120 gasometer images for training, validation and testing of 3 types of deep neural network. 22,985 images were used for training and validation, 4,135 images were used for testing. We randomly splitted 22,985 images with 8:2 ratio for training and validation respectively for each training epoch. 4,135 test image were categorized into 5 types (Normal, noise, reflex, scale and slant). Normal data is clean image data, noise means image with noise signal, relfex means image with light reflection in gasometer region, scale means images with small object size due to long-distance capturing and slant means images which is not horizontally flat. Final character string recognition accuracies for device ID and gas usage amount of normal data are 0.960 and 0.864 respectively.

• Journal of Energy Engineering
• /
• v.23 no.1
• /
• pp.21-32
• /
• 2014

This study has analyzed whether the virtuous cycle of value-added between the processes within the company has formed and whether the virtuous ecosystem between the processes within the industry has been built through the analysis of value chain(VC) and value system(VS) targeting the Korean photovoltaic companies. For a study method, after conducting a survey on the companies, a regression analysis was performed on the causal relationship between the process within the VC and VS. Based on the results of the analysis, for the VC of the Korean photovoltaic industry, an increase in the R&D support from the government has led to the increase in the investment of R&D for the related industry, and the increase in the investment of R&D has contributed to the increase in the growth of its productivity, and the growth in the productivity of R&D has influenced the increase in the production of solar products. In addition, the reduction of photovoltaic production cost for the company has influenced the increase of recurring profit margin compared to the sales. However it was shown that the increase in the company's production volume does not contribute to the reduction of production cost. Meanwhile, the increase in recurring profit margin compared to the sales were influencing the increase in the production volume but it was shown that the increase in the company's investment of R&D was not a contributing factor thus it was not included in the virtuous cycle. It was analyzed that the VS was shown not to influence all other processes within the industry except for the module companies where the increase in the recurring profit margin compared to the sales was influenced by the increase in the recurring profit margin compared to the sales of solar cell companies. This shows that the virtuous industrial ecosystem which should be made under the mutual cooperation by the ingot, wafer, solar cell, module and system companies are yet incomplete.

• Journal of Korean Society of Transportation
• /
• v.22 no.7 s.78
• /
• pp.119-129
• /
• 2004

This paper is designed to report the results of development and validation procedures in relation to the Freeway Incident Management System (FIMS) prototype development as part of Intelligent Transportation Systems Research and Development program. The central core of the FIMS is an integration of the component parts and the modular, but the integrated system for freeway management. The whole approach has been component-orientated, with a secondary emphasis being placed on the traffic characteristics at the sites. The first action taken during the development process was the selection of the required data for each components within the existing infrastructure of Korean freeway system. After through review and analysis of vehicle detection data, the pilot site led to the utilization of different technologies in relation to the specific needs and character of the implementation. This meant that the existing system was tested in a different configuration at different sections of freeway, thereby increasing the validity and scope of the overall findings. The incident detection module has been performed according to predefined system validation specifications. The system validation specifications have identified two component data collection and analysis patterns which were outlined in the validation specifications; the on-line and off-line testing procedural frameworks. The off-line testing was achieved using asynchronous analysis, commonly in conjunction with simulation of device input data to take full advantage of the opportunity to test and calibrate the incident detection algorithms focused on APID, DES, DELOS and McMaster. The simulation was done with the use of synchronous analysis, thereby providing a means for testing the incident detection module.