• 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.

• Korean Journal of Environment and Ecology
• /
• v.36 no.3
• /
• pp.220-256
• /
• 2022

The most important thing for conserving plant diversity in an area is to make an overall inventory of the plant species inhabiting the area. In particular, limestone areas are known for their high plant diversity and distribution of specific plants. Despite that, only a few have been designated as protected areas. This study investigated the vascular plants distributed in Goyangsan Mtn. and Munraesan Mtn., located in limestone areas of the central part of the Korean Peninsula. A field survey was conducted eight times from April to October 2021. As a result, we identified a total of 654 taxa comprising 113 families, 357 genera, 592 species, 15 subspecies, 44 varieties, and 3 formulas. They included four endangered wild plant species: Astilboides tabularis, Eleutherococcus senticosus, Cypripedium macranthos, and Epilobium hirsutum. Endemic plants in Korea were identified as 32 taxa. Floristic target plants were identified as 168 taxa, specifically 5 taxa of grade V, 41 taxa of grade IV, and 36 taxa of grade III. The red data plants included 2 taxa as "Endangered (EN)", 7 taxa as "Vulnerable (VU)", and 7 taxa as "Near threatened (NT)". A total of 41 taxa of naturalized plants were identified, and 4 of them were invasive alien plants. The surveyed vicinity of Goyangsan Mtn. and Munraesan Mtn. showed high plant diversity and contained core habitats for distribution of an endangered wild plant, Astilboides tabularis,in the limestone area. Moreover, both mountains contained a small population of Cotoneaster integerrimus. These findings confirm that the area has conservation values. Therefore, we propose to identify areas with high plant diversity and designate them as special protected areas.

• 지역과문화
• /
• v.7 no.1
• /
• pp.59-98
• /
• 2020

An empirical study utilizing data was performed with a variable called 'strategic planning' for festivals in order to look into the impact of a cultural tourism festival on urban regeneration. As a success factor of a festival, strategic festival planning was drawn up, and the following hypotheses were set: Seven strategic factors verified through an exploratory factor analysis will have a positive impact on festival satisfaction (festival success) and on urban regeneration, and festival satisfaction will have a positive impact on urban regeneration by having a mediating effect on it. For the analysis, the Gimje Horizon Festival was selected as it was considered as a typical case of urban regeneration through a festival, and the relationship between the festival and urban regeneration was understood by conducting a combined analysis of a quantitative analysis through a survey, a literature search, field investigations and in-depth interviews. The quantitative analysis indicates that strategic planning has a positive impact on festival satisfaction (festival success) and on urban regeneration and that festival satisfaction has a positive impact on urban regeneration. The same study result as the quantitative analysis result was obtained even through a qualitative analysis. This shows that the higher the path coefficient of strategic planning, the higher the path coefficient of festival satisfaction and urban generation and that with better strategic planning, the effects of festival satisfaction and urban regeneration are maximized. In other words, when planning and implementing a festival by actively incorporating the seven strategic planning factors which were suggested as festival success factors earlier in this study beginning from the stage of festival planning, it is likely to have a positive impact not only on the success of the festival but also on urban regeneration. This is an implication that gives a new alternative to software-based urban regeneration through festivals. It is meaningful to present the importance of festival planning and the direction of planning to maximize the effect of urban regeneration to festival planners and urban regeneration experts. This study is believed to serve as a momentum for people to take a new approach to studies on festivals and urban regeneration utilizing software in the future.

• Journal of Intelligence and Information Systems
• /
• v.28 no.4
• /
• pp.251-273
• /
• 2022

Recently, research on applying text analysis to deep learning has steadily continued. In particular, researches have been actively conducted to understand the meaning of words and perform tasks such as summarization and sentiment classification through a pre-trained language model that learns large datasets. However, existing pre-trained language models show limitations in that they do not understand specific domains well. Therefore, in recent years, the flow of research has shifted toward creating a language model specialized for a particular domain. Domain-specific pre-trained language models allow the model to understand the knowledge of a particular domain better and reveal performance improvements on various tasks in the field. However, domain-specific further pre-training is expensive to acquire corpus data of the target domain. Furthermore, many cases have reported that performance improvement after further pre-training is insignificant in some domains. As such, it is difficult to decide to develop a domain-specific pre-trained language model, while it is not clear whether the performance will be improved dramatically. In this paper, we present a way to proactively check the expected performance improvement by further pre-training in a domain before actually performing further pre-training. Specifically, after selecting three domains, we measured the increase in classification accuracy through further pre-training in each domain. We also developed and presented new indicators to estimate the specificity of the domain based on the normalized frequency of the keywords used in each domain. Finally, we conducted classification using a pre-trained language model and a domain-specific pre-trained language model of three domains. As a result, we confirmed that the higher the domain specificity index, the higher the performance improvement through further pre-training.

• Economic and Environmental Geology
• /
• v.55 no.4
• /
• pp.407-419
• /
• 2022

Soil(vadose zone) gas compositions were measured for about 3 days to suggest a method for monitoring and interpreting soil gas data collected around wells from which methane(CH₄) is outflowing. The vadose zone gas samples were collected within 1 m around two test wells(TB2 and TB3) at Pohang and analyzed for CO₂, CH₄, N₂ and O₂ concentrations in situ. CO₂ flux was measured beside TB2. In addition, gas samples from well head in TB2 and atmospheric air samples were collected for comparison. Carbon isotopes of CO₂(δ¹³C_CO2) of samples collected on the last day of the study period were analyzed in the laboratory. The two test wells (TB2 and 3) were 12.7 m apart and only TB3 was cemented to the surface. According to the bio-geochemical process-based interpretation, the relationships between CO₂ and O₂, N₂, and N₂/O₂ of vadose zone gas were plotted between the lines of CH₄ oxidation and CO₂ dissolution. In addition, the CH₄ concentrations of gas samples from the wellhead of the uncemented well (TB2) were 5.2 times higher than the atmospheric CH₄ concentration. High CO₂ concentrations (average 1.148%) of vadose zone gas around TB2 seemed to be attributed to the oxidation of CH₄. On the other hand, the vadose zone CO₂ around the cemented well(TB3) showed a relatively low concentration(0.136%). This difference indicates that the vadose zone gas(including CO₂) around the CH₄ outflowing well were strongly affected by well completion(cementing). This study result can be used to establish strategies for environmental monitoring of soil around natural gas sites, and can be used to monitor leakage around injection and observation wells for CO₂ geological storage. In addition, the method of this study is useful for soil monitoring in natural gas storage and oil-contaminated sites.

• Journal of the Korea Organic Resources Recycling Association
• /
• v.29 no.4
• /
• pp.67-76
• /
• 2021

Biogasification is a technology that produces environmentally friendly fuel using methane gas generated in the process of stably decomposing and processing organic waste. Biogasification is the most used method for energy conversion of organic waste with high moisture content, and is a useful method for organic waste treatment following the prohibition of direct landfill (2005) and marine dumping (2013). Due to African Swine Fever (ASF), which recently occurred in Korea, recycling of wet feed is prohibited, and consumers such as dry feed and compost are negatively recognized, making it difficult to treat food waste. Accordingly, biogasification is attracting more attention for the treatment and recycling of food waste. Korea's energy consumption amounted to 268.41 106toe, ranking 9th in the world. However, it is an energy-poor country that depends on foreign imports for about 95.8% of its energy supply. Therefore, in Korea, the Renewable Energy Portfolio Standard (RPS) is being introduced. The domestic RPS system sets the weight of the new and renewable energy certificate (REC, Renewable energy certificate) of waste energy lower than that of other renewable energy. Therefore, an additional incentive system is required for the activation of waste-to-energy. In this study, the operation of an anaerobic digester that treats food waste, food waste Leachate and various organic wastes was confirmed. It was intended to be used as basic data for preparing the waste-to-energy incentive system through precise monitoring for a certain period of time. Four sites that produce biogas from organic waste and use them for power generation and heavy gas were selected as target facilities, and field surveys and sampling were conducted. Basic properties analysis was performed on the influent sample of organic waste and the effluent sample according to the treatment process. As a result of the analysis of the properties, the total solids of the digester influent was an average of 12.11%, and the volatile solids of the total solids were confirmed to be 85.86%. BOD and CODcr removal rates were 60.8% and 64.8%. The volatile fatty acids in the influent averaged 55,716 mg/L. It can be confirmed that most of the volatile fatty acids were decomposed and removed with an average reduction rate of 92.3% after anaerobic digestion.

• Economic and Environmental Geology
• /
• v.55 no.5
• /
• pp.551-561
• /
• 2022

Geophysical exploration methods are very useful for generating high-resolution images of underground structures, and such methods can be applied to investigation of buried cultural properties and for determining their exact locations. In this study, image feature extraction and image segmentation methods were applied to automatically distinguish the structures of buried relics from the high-resolution ground-penetrating radar (GPR) images obtained at the center of Silla Kingdom, Gyeongju, South Korea. The major purpose for image feature extraction analyses is identifying the circular features from building remains and the linear features from ancient roads and fences. Feature extraction is implemented by applying the Canny edge detection and Hough transform algorithms. We applied the Hough transforms to the edge image resulted from the Canny algorithm in order to determine the locations the target features. However, the Hough transform requires different parameter settings for each survey sector. As for image segmentation, we applied the connected element labeling algorithm and object-based image analysis using Orfeo Toolbox (OTB) in QGIS. The connected components labeled image shows the signals associated with the target buried relics are effectively connected and labeled. However, we often find multiple labels are assigned to a single structure on the given GPR data. Object-based image analysis was conducted by using a Large-Scale Mean-Shift (LSMS) image segmentation. In this analysis, a vector layer containing pixel values for each segmented polygon was estimated first and then used to build a train-validation dataset by assigning the polygons to one class associated with the buried relics and another class for the background field. With the Random Forest Classifier, we find that the polygons on the LSMS image segmentation layer can be successfully classified into the polygons of the buried relics and those of the background. Thus, we propose that these automatic classification methods applied to the GPR images of buried cultural heritage in this study can be useful to obtain consistent analyses results for planning excavation processes.

• Journal of the Korean Institute of Landscape Architecture
• /
• v.50 no.5
• /
• pp.40-53
• /
• 2022

This study carried out to provide the methodology and basic status material of using Korean national statistics needed to find the actual state of the landscape architecture industry. The landscape architecture industry was classified into 'Design', 'Construction Management', 'construction', 'Maintenance & Management', 'Materials', 'Research', 'Education', and 'Administration' areas. In each field, business types were systemized and associated in accordance with Korean standard industrial classification and legislations pertinent to construction. Among them, the business types directly defined in the construction related legislations under the Ministry of Land, Infrastructure and Transport were focused on, and the establishment, association, integration, distribution, duplication, and omission of national statistics were analyzed. As a result, the business types of statistical analysis were selected. In order for commonality of statistical items and minimized error of interpretation, semantic analysis was conducted. Finally, the number of registered business types, the number of workers, and sales were selected. Based on them, the analysis framework applicable to fundamental analysis and evaluation of the actual state of the industry was proposed. Actual national statical data were applied for analysis and evaluation. In 2019, the number of registered business types related to the landscape architecture industry was 12,160, the number of workers by business type was 106,296, and the sales by business type were 8,308.5 billion KRW. The number of registered business types and the number of workers had been on the rise from 2017, whereas the sales had been on the decrease. It is required to come up with a plan for industrial development. This study was conducted with the national statistics established by multiple public institutions, so that there are limitations in securing consistency and reliability. Therefore, it is necessary to establish systematic and consistent national statistics in accordance with 「Landscaping Promotion Act」. In the future, it will planned to research application and development plans of national statistics according to subjects including park and green.

• Journal of Life Science
• /
• v.33 no.1
• /
• pp.56-63
• /
• 2023

In this study, two cases of food poisoning caused by Salmonella that occurred in Gyeongsangnam-do in September 2021 are reported. One of the outbreaks occurred in a school and the other in a company. The molecular epidemiological characteristics of the isolated strains in the two outbreaks were analyzed. In the case of the school outbreak, 29 (4.9%) of 588 individuals experienced diarrhea and abdominal pain. As a result of a test of 36 individuals (patients, n=29; cook workers, n=7), Salmonella enterica serotype Enteritidis was detected in 17 (47.2%) patients, suggesting this serotype was the principal cause. Meanwhile, Salmonella spp. were not detected in 35 food and environmental samples. In the company outbreak, 87 (3.0%) of 2,900 individuals who had intaked from the same source experienced diarrhea, abdominal pain, and fever. In a test of 50 individuals (patients, n=40; cook workers, n=10), S. Enteritidis was detected in 28 patients (56.0%). Also, Vibrio cholerae (NAG) was detected in four patients with S. Enteritidis, and V. cholerae (NAG) only was detected in one patient. Salmonella spp. were not detected in 118 preserved foods, but S. Enteritidis was detected in one eaten food (toast) delivered in group by the company. Through PFGE genetic homology analysis of the isolated strains, all S. Enteritidis detected in patients and consumed foods were the same type. It seems that these S. Enteritidis isolates were the same type as detected in a previous school outbreak and in patients of group food poisoning in other regions, leading to an enhanced problem of food poisoning and epidemiology. Our analytic results can provide data for epidemiological management and food poisoning prevention based on molecular characteristics.

• Journal of the Korean Society of Radiology
• /
• v.17 no.4
• /
• pp.489-495
• /
• 2023

Radiation shielding facilities are constructed in locations where diagnostic radiation generators are installed, with the aim of preventing exposure for patients and radiation workers. The purpose of this study is seek to compare and validate the trend of attenuation thickness of lead, the primary material in these radiation shielding facilities, at different maximum tube voltages by Monte Carlo simulations and measurement. We employed the Monte Carlo N-Particle 6 simulation code. Within this simulation, we set a lead shielding arrangement, where the distance between the source and the lead sheet was set at 100 cm and the field of view was set at 10 × 10 cm². Additionally, we varied the tube voltages to encompass 80, 100, 120, and 140 kVp. We calculated energy spectra for each respective tube voltage and applied them in the simulations. Lead thicknesses corresponding to attenuation rates of 50, 70, 90, and 95% were determined for tube voltages of 80, 100, 120, and 140 kVp. For 80 kVp, the calculated thicknesses for these attenuation rates were 0.03, 0.08, 0.21, and 0.33 mm, respectively. For 100 kVp, the values were 0.05, 0.12, 0.30, and 0.50 mm. Similarly, for 120 kVp, they were 0.06, 0.14, 0.38, and 0.56 mm. Lastly, at 140 kVp, the corresponding thicknesses were 0.08, 0.16, 0.42, and 0.61 mm. Measurements were conducted to validate the calculated lead thicknesses. The radiation generator employed was the GE Healthcare Discovery XR 656, and the dosimeter used was the IBA MagicMax. The experimental results showed that at 80 kVp, the attenuation rates for different thicknesses were 43.56, 70.33, 89.85, and 93.05%, respectively. Similarly, at 100 kVp, the rates were 52.49, 72.26, 86.31, and 92.17%. For 120 kVp, the attenuation rates were 48.26, 71.18, 87.30, and 91.56%. Lastly, at 140 kVp, they were measured 50.45, 68.75, 89.95, and 91.65%. Upon comparing the simulation and experimental results, it was confirmed that the differences between the two values were within an average of approximately 3%. These research findings serve to validate the reliability of Monte Carlo simulations and could be employed as fundamental data for future radiation shielding facility construction.