• Journal of Intelligence and Information Systems
• /
• v.28 no.2
• /
• pp.191-206
• /
• 2022

Recently, it is a de-facto approach to utilize a pre-trained language model(PLM) to achieve the state-of-the-art performance for various natural language tasks(called downstream tasks) such as sentiment analysis and question answering. However, similar to any other machine learning method, PLM tends to depend on the data distribution seen during the training phase and shows worse performance on the unseen (Out-of-Distribution) domain. Due to the aforementioned reason, there have been many efforts to develop domain-specified PLM for various fields such as medical and legal industries. In this paper, we discuss the training of a finance domain-specified PLM for the Korean language and its applications. Our finance domain-specified PLM, KB-BERT, is trained on a carefully curated financial corpus that includes domain-specific documents such as financial reports. We provide extensive performance evaluation results on three natural language tasks, topic classification, sentiment analysis, and question answering. Compared to the state-of-the-art Korean PLM models such as KoELECTRA and KLUE-RoBERTa, KB-BERT shows comparable performance on general datasets based on common corpora like Wikipedia and news articles. Moreover, KB-BERT outperforms compared models on finance domain datasets that require finance-specific knowledge to solve given problems.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.67 no.1
• /
• pp.27-40
• /
• 2022

Building self-sustainable rural infrastructure and environment through smart digital agriculture technology innovation is one of the major goals of the Korean agricultural administration as a part of the nation's 4^th industry revolution. To identify areas for improving and effectively investing in the acceleration of rural development, 207 experts in the areas of crop science and smart digital agriculture technology were interviewed for their opinions and suggestions on 22 questions designed to recognize fundamental agricultural issues to be addressed and solutions to advance technology innovation and rural development. Majority of the participants expected smart digital agriculture technologies to resolve major agricultural issues and help build a better rural environment. To overcome technology gaps and resolve issues more effectively, further investment in training new technology experts and building stronger agricultural technology infrastructure is urgent, and persistent and systematic support from agricultural administration appears to be the key for accelerating the process. While the leading global groups of both public and private sectors have advanced their technologies beyond the field application stage, most of the Korean technologies remain at the early pilot stage. Aging population and lack of labor in rural areas, unknown future climate change, and challenges in sustainable rural development are expected to be resolved by smart digital agriculture technologies. Technological innovations by research institutes should be promptly deployed in the crop production field, and farm training systemically organized by local technology centers can accelerate farming revolution. Standardization of equipment and data systems is another key to the success of digitalization of food crop production and food supply chains nationwide.

• The Journal of Engineering Geology
• /
• v.31 no.4
• /
• pp.603-620
• /
• 2021

A land nodal seismic system was employed to acquire seismic reflection data using stand-alone cable-free receivers in a land-river area. Acquiring reliable data using this technology is very cost effective, as it avoids topographic problems in the deployment and collection of receivers. The land nodal airgun system deployed on the mouth of the Hyungsan River (in Pohang, Gyeongsangbuk Province) used airgun sources in the river and receivers on the riverbank, with subparallel source and receiver lines, approximately 120 m-spaced. Seismic data collected on the riverbank are characterized by a low signal-to-noise (S/N) and inconsistent reflection events. Most of the events are represented by hyperbola in the field records, including direct waves, guided waves, air waves, and Scholte surface waves, in contrast to the straight lines in the data collected conventionally where source and receiver lines are coincident. The processing strategy included enhancing the signal behind the low-frequency large-amplitude noise with a cascaded application of bandpass and f-k filters for the attenuation of air waves. Static time delays caused by the cross-offset distance between sources and receivers are corrected, with a focus on mapping the shallow reflections obscured by guided wave and air wave noise. A new time-distance equation and curve for direct and air waves are suggested for the correction of the static time delay caused by the cross-offset between source and receiver. Investigation of the minimum cross-offset gathers shows well-aligned shallow reflections around 200 ms after time-shift correction. This time-delay static correction based on the direct wave is found essential to improving the data from parallel source and receiver lines. Data acquisition and processing strategies developed in this study for land nodal airgun seismic systems will be readily applicable to seismic data from land-sea areas when high-resolution signal data becomes available in the future for investigation of shallow gas reservoirs, faults, and engineering designs for the development of coastal areas.

• Journal of Korean Society of Disaster and Security
• /
• v.14 no.4
• /
• pp.69-80
• /
• 2021

This study tried to suggest improvement measures by discovering problems or matters requiring improvement among the annual regional safety evaluation systems. Briefly introducing the structure and contents of the study, which is the introduction, describes the regional safety evaluation method newly applied by the Ministry of Public Administration and Security in 2020. Utilization plans were also introduced according to the local safety level that was finally evaluated by the local government. In this paper, various views of previous researchers related to regional safety are summarized and described. In addition, problems were drawn in the composition of the index of local safety, the method of calculating the index, and the application of the current index. Next, the problems of specific regional safety evaluation indicators were analyzed and solutions were presented. First, "Number of semi-basement households" is replaced with "Number of households receiving basic livelihood" of 「Social Vulnerability Index」 in the field of disaster risk factors is replaced with "the number of households receiving basic livelihood". In addition, the "Vinyl House Area" is evaluated by replacing "the number of households living in a Vinyl House, the number of container households, and the number of households in Jjok-bang villages" with data. Second, in the management and evaluation of habitual drought disaster areas, local governments with a water supply rate of 95% or higher in Counties, Cities, and Districts are treated as "missing". This is because drought disasters rarely occur in the metropolitan area and local governments that have undergone urbanization. Third, the activities of safety sheriffs, safety monitor volunteers, and disaster safety silver monitoring groups along with the local autonomous prevention foundation are added to the evaluation of the evaluation index of 「Regional Autonomous Prevention Foundation Activation」 in the field of response to disaster prevention measures. However, since the name of the local autonomous disaster prevention organization may be different for each local government, if it is an autonomous disaster prevention organization organized and active for disaster prevention, it would be appropriate to evaluate the results by summing up all of its activities. Fourth, among the Scorecard evaluation items, which is a safe city evaluation tool used by the United Nations Office for Disaster Risk Reduction(UNDRR), the item "preservation of natural buffers to strengthen the protection functions provided by natural ecosystems" is borrowed, which is closely related to natural disasters. The Scorecard evaluation is an assessment index that focuses on improving the disaster resilience of local governments while carrying out the campaign "Creating cities resilient to climate crises and disasters" emphasized by UNDRR. Finally, the names of "regional safety level" and "local safety index" are similar, so the term of local safety level is changed to "natural disaster safety level" or "natural calamity safety level". This is because only the general public can distinguish the local safety level from the local safety index.

• Journal of Intelligence and Information Systems
• /
• v.28 no.2
• /
• pp.127-146
• /
• 2022

Recently, as word embedding has shown excellent performance in various tasks of deep learning-based natural language processing, researches on the advancement and application of word, sentence, and document embedding are being actively conducted. Among them, cross-language transfer, which enables semantic exchange between different languages, is growing simultaneously with the development of embedding models. Academia's interests in vector alignment are growing with the expectation that it can be applied to various embedding-based analysis. In particular, vector alignment is expected to be applied to mapping between specialized domains and generalized domains. In other words, it is expected that it will be possible to map the vocabulary of specialized fields such as R&D, medicine, and law into the space of the pre-trained language model learned with huge volume of general-purpose documents, or provide a clue for mapping vocabulary between mutually different specialized fields. However, since linear-based vector alignment which has been mainly studied in academia basically assumes statistical linearity, it tends to simplify the vector space. This essentially assumes that different types of vector spaces are geometrically similar, which yields a limitation that it causes inevitable distortion in the alignment process. To overcome this limitation, we propose a deep learning-based vector alignment methodology that effectively learns the nonlinearity of data. The proposed methodology consists of sequential learning of a skip-connected autoencoder and a regression model to align the specialized word embedding expressed in each space to the general embedding space. Finally, through the inference of the two trained models, the specialized vocabulary can be aligned in the general space. To verify the performance of the proposed methodology, an experiment was performed on a total of 77,578 documents in the field of 'health care' among national R&D tasks performed from 2011 to 2020. As a result, it was confirmed that the proposed methodology showed superior performance in terms of cosine similarity compared to the existing linear vector alignment.

• Journal of Intelligence and Information Systems
• /
• v.28 no.1
• /
• pp.329-352
• /
• 2022

It is reported that particulate matter(PM) penetrates the lungs and blood vessels and causes various heart diseases and respiratory diseases such as lung cancer. The subway is a means of transportation used by an average of 10 million people a day, and although it is important to create a clean and comfortable environment, the level of particulate matter pollution is shown to be high. It is because the subways run through an underground tunnel and the particulate matter trapped in the tunnel moves to the underground station due to the train wind. The Ministry of Environment and the Seoul Metropolitan Government are making various efforts to reduce PM concentration by establishing measures to improve air quality at underground stations. The smart air quality management system is a system that manages air quality in advance by collecting air quality data, analyzing and predicting the PM concentration. The prediction model of the PM concentration is an important component of this system. Various studies on time series data prediction are being conducted, but in relation to the PM prediction in subway stations, it is limited to statistical or recurrent neural network-based deep learning model researches. Therefore, in this study, we propose four transformer-based models including spatiotemporal transformers. As a result of performing PM concentration prediction experiments in the waiting rooms of subway stations in Seoul, it was confirmed that the performance of the transformer-based models was superior to that of the existing ARIMA, LSTM, and Seq2Seq models. Among the transformer-based models, the performance of the spatiotemporal transformers was the best. The smart air quality management system operated through data-based prediction becomes more effective and energy efficient as the accuracy of PM prediction improves. The results of this study are expected to contribute to the efficient operation of the smart air quality management system.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.68 no.3
• /
• pp.121-133
• /
• 2023

The purpose of this study was to improve weed management systems under varying carbon dioxide concentrations and temperatures by evaluating the growth of Acalypha australis and observing the efficacy of four foliar and four soil herbicides, as well as measuring phytotoxicity in soybean crops treated with these herbicides. In both growth chamber and greenhouse conditions, plant height and shoot fresh weight of Acalypha australis increased as temperature increased. The variable to maximum fluorescence ratio (Fv/Fm), relative electron transport rate (ETR), plant height, leaf area, and shoot fresh weight of Acalypha australis were higher at carbon dioxide concentrations of 800 ppm than at 400 ppm. The efficacy of a foliar herbicide, glufosinate, on Acalypha australis was lower at 30℃ than at 20℃ and 25℃ in the growth chamber condition and was also lower at 29℃ than at 21℃ and 25℃ in greenhouse conditions. In contrast, mecoprop efficacy on Acalypha australis was lower at 20℃ and 25℃ than at 30℃ in growth chamber conditions and lower at 21℃ and 25℃ than at 29℃ in greenhouse conditions. Glyphosate efficacy was lower at 21℃ than at 25℃ and 29℃ under greenhouse conditions. With soil herbicides, metolachlor and ethalfluraline, efficacies were higher at relatively high temperatures under both growth chamber and greenhouse conditions. However, in the case of linuron, the difference in efficacy was not observed under varying temperatures in both growth chamber and greenhouse conditions. When ¼ of the recommended glyphosate rates were applied to Acalypha australis, efficacy was lower under 800 ppm carbon dioxide concentrations than under 400 ppm. In contrast, when ¼ of the recommended rate of bentazone was applied to Acalypha australis, efficacy was higher under 800 ppm carbon dioxide concentrations than under 400 ppm. Despite application rates, glufosinate efficacy differed insignificantly under different carbon dioxide concentrations. When applied at ¼ of the recommended rate, the efficacy of ethalfuralin was higher under 800 ppm carbon dioxide concentrations than under 400 ppm. However, efficacies of other herbicides were not different despite varying carbon dioxide concentrations. Soybean phytotoxicity in crops treated with the recommended rate and twice the recommended rate of soil herbicides was not significantly different regardless of temperature and carbon dioxide concentrations. Overall, weed efficacy of some herbicides decreased in response to different temperatures and carbon dioxide concentrations. Therefore, new weed management methods are required to ensure high rates of weed control in conditions affected by climate change.

• Korean Journal of Clinical Laboratory Science
• /
• v.56 no.2
• /
• pp.147-155
• /
• 2024

Gallium-68-prostate-specific membrane antigen-11 (⁶⁸Ga-PSMA-11) is a positron emission tomography radiopharmaceutical that labels a Glu-urea-Lys-based ligand with ⁶⁸Ga, binding specifically to the PSMA. It is used widely for imaging recurrent prostate cancer and metastases. On the other hand, the preparation and quality control testing of ⁶⁸Ga-PSMA-11 in medical institutions takes over 60 minutes, limiting the daily capacity of ⁶⁸Ge/⁶⁸Ga generators. While the generator provides 1,110 MBq (30 mCi) nominally, its activity decreases over time, and the labeling yield declines irregularly. Consequently, additional preparations are needed, increasing radiation exposure for medical technicians, prolonging patient wait times, and necessitating production schedule adjustments. This study aimed to reduce the ⁶⁸Ga-PSMA-11 preparation time and optimize the automated synthesis system. By shortening the reaction time between ⁶⁸Ga and the PSMA-11 precursor and adjusting the number of purification steps, a faster and more cost-effective method was tested while maintaining quality. The final synthesis time was reduced from 30 to 20 minutes, meeting the standards for the HEPES content, residual solvent EtOH content, and radiochemical purity. This optimized procedure minimizes radiation exposure for medical technicians, reduces patient wait times, and maintains consistent production schedules, making it suitable for clinical application.

• The Korean Society of Law and Medicine
• /
• v.24 no.3
• /
• pp.155-206
• /
• 2023

When a person with diminished mental capacity refuses necessary medical care, normative judgments about when paternalistic intervention can be justified come into question. A typical example is involuntary hospitalization for people with mental disabilities, traditionally governed by mental health law. However, Korean civil law reform in 2011 introduced a new form of involuntary hospitalization through guardianship legislation, leading to a dualized system to involuntary hospitalization. Consequently, a conflict has arisen between the 'best interest and surrogate decision-making' paradigm of civil law and the 'social defense and preventive detention' paradigm of mental health law. Many countries have criticized this dualized system as not only inefficient but also unfair. Moreover, the requirement for the presence of 'mental illness' for involuntary hospitalization under mental health law has faced criticism for unfairly discriminating against people with mental disabilities. In response, attempts have been made to integrate guardianship legislation and mental health law based on mental capacity. This study examines the legislative process and framework of the Mental Capacity Act (Northern Ireland) 2016, which reorganized the mental health care system by fusing guardianship legislation with mental health law based on mental capacity. By analyzing the case of Northern Ireland, which has grappled with conflicts between guardianship legislation and mental health law since the 1990s and recently proposed mental capacity as a single, non-discriminatory standard, we aimed to offer insights for the Korean guardianship and mental health systems.

• Journal of the Korean Society of Radiology
• /
• v.18 no.3
• /
• pp.187-201
• /
• 2024

Osteoporosis is a major health issue globally, often remaining undetected until a fracture occurs. To facilitate early detection, deep learning (DL) models were developed to classify osteoporosis using abdominal computed tomography (CT) scans. This study was conducted using retrospectively collected data from 3,012 contrast-enhanced abdominal CT scans. The DL models developed in this study were constructed for using image data, demographic/clinical information, and multi-modality data, respectively. Patients were categorized into the normal, osteopenia, and osteoporosis groups based on their T-scores, obtained from dual-energy X-ray absorptiometry, into normal, osteopenia, and osteoporosis groups. The models showed high accuracy and effectiveness, with the combined data model performing the best, achieving an area under the receiver operating characteristic curve of 0.94 and an accuracy of 0.80. The image-based model also performed well, while the demographic data model had lower accuracy and effectiveness. In addition, the DL model was interpreted by gradient-weighted class activation mapping (Grad-CAM) to highlight clinically relevant features in the images, revealing the femoral neck as a common site for fractures. The study shows that DL can accurately identify osteoporosis stages from clinical data, indicating the potential of abdominal CT scans in early osteoporosis detection and reducing fracture risks with prompt treatment.