• The Bulletin of The Korean Astronomical Society
• /
• v.40 no.1
• /
• pp.63.3-64
• /
• 2015

In our previous work (Kim et al., 2015), we suggested a new classification scheme, which categorizes the SPEs into four groups based on association with flare or CME inferred from onset timings as well as proton acceleration patterns using multienergy observations. In this study, we have tried to find whether there are any typical characteristics of associated events and acceleration sites in each group using 42 SPEs from 1997 to 2012. We find: (i) if the proton acceleration starts from a lower energy, a SPE has a higher chance to be a strong event (>5000pfu) even if the associated flare and CME are not so strong. The only difference between the SPEs associated with flare and CME is the location of the acceleration site. For the former, the sites are very low (~1Rs) and close to the western limb, while the latter has a relatively higher and wider acceleration sites. (ii) When the proton acceleration starts from the higher energy, a SPE tends to be a relatively weak event (<1000pfu), in spite of its associated CME is relatively stronger than previous group. (iii) The SPEs categorized by the simultaneous proton acceleration in whole energy range within 10 minutes, tend to show the weakest proton flux in spite of strong related eruptions. Their acceleration heights are very close to the locations of type II radio bursts. Based on those results, we suggest that the different characteristics of the four groups are mainly due to the different mechanisms governing the acceleration pattern and interval, and different condition such as the acceleration location.

• Journal of Korea Water Resources Association
• /
• v.43 no.7
• /
• pp.597-608
• /
• 2010

In this study, a method for soil moisture estimation was proposed to obtain the nationwide soil moisture distribution map using on-site soil moisture observations, rainfall, surface temperature, NDVI, land cover, effective soil depth, and CART (Classification And Regression Tree) algorithm. The method was applied to the Yong-dam dam basin since the soil moisture data (4 sites) of the basin were reliable. Soil moisture observations of 3 sites (Bu-gui, San-jeon, Cheon-cheon2) were used for training the algorithm and 1 site (Gye-buk2) was used for the algorithm validation. The correlation coefficient between the observed and estimated data of soil moisture in the validation sites is about 0.737. Results show that even though there are limitations of the lack of reliable soil moisture observation for various land use, soil type, and topographic conditions, the soil moisture estimation method using ancillary data and CART algorithm can be a reasonable approach since the algorithm provided a fairly good estimation of soil moisture distribution for the study area.

• Maxillofacial Plastic and Reconstructive Surgery
• /
• v.31 no.1
• /
• pp.77-80
• /
• 2009

Mucosa-associated lymphoid tissue(MALT) lymphoma is thought to originate from marginal zone B-cells. In the WHO classification, Extranodal marginal zone lymphoma of MALT is classified B-cell Non-Hodgkin lymphoma. Common sites of MALT lymphoma include stomach, lung and the ocular-adnexa. Although less common in other sites, it is the most common low-grade lymphoma of the breast, thyroid, bowel skin and soft tissue. No strong age or gender predominance exists in MALT lymphoma. Dissemination to other sites can occur. In the oral cavity, MALT lymphoma is rare. Herein, we present a case of intra-oral MALT lymphoma. 66 year-old woman without any background of immunodeficiency or autoimmune disease admitted department of oral & maxillofacial surgery in Ulsan university hospital for evaluation of long-standing mild upper lip swelling. The lesion was completely resected and biopsied. Histological and immunohistochemical stains(CD3, CD5, CD20, CD21, CK) findings were used to confirm the lesion. Bone marrow biopsy was done and no bone marrow involvement was found. She did not receive chemotherapy and radiotherapy after surgery. No recurrence has been noted in the 22 months to date.

• KIEAE Journal
• /
• v.10 no.1
• /
• pp.3-10
• /
• 2010

Riverside forests make a river bank stable because trees of them hold together the stone and soil by roots and decrease the speed of running fluid by trunks. So they become known to have positive effects on flood prevention. So This study will be a basic study to preserve and restore of riverside forests. The goal of this study is to classify types of Han riverside forests between Yangpyeng and Yeoju gun. and find out sites of planting. Results of this study can be summarized as follows; The evaluation indicators were set up based on literature review and site survey. Two indicator categories were developed: natural environment and human environment. And they were divided into 5 sub-categories for calculating weights. As for the major indicator categories, the weighted index of natural environment is at 0.5. And the weighted index of human environment is at 0.5 followed by access at 0.15, the range of user at 0.15, cultivated land at 0.1 and legislation at 0.1. This study selected 53 sites for riverside forests planting. They were classified with types of bank(11), level-upped riverside(32), island(10). The amount of the length of 11 bank types is 23,050m, the area of 32 level-upped riverside types is $4,490,000m^2$ and the area of 10 island types is $4,590,000m^2$. After the evaluation of 53 riverside forests, this study selected 12 sites of riverside forests. They were two bank types, nine level-upped riverside types, and one island type. Rebuilding riverside forests are to accomplish the green network which links and divides region. It will be one of the best ecological methods to construct friendly environmental region.

• The KIPS Transactions:PartA
• /
• v.19A no.4
• /
• pp.187-194
• /
• 2012

This paper proposes a string analysis based system for classifying Android Apps that may access so called harmful sites, and shows an experiment result for real Android apps on the market. The system first transforms Android App binary codes into Java byte codes, it performs string analysis to compute a set of strings at all program points, and it classifies the Android App as bad ones if the computed set contains URLs that are classified because the sites provide inappropriate contents. In the proposed approach, the system performs such a classification in the stage of distribution before installing and executing the Apps. Furthermore, the system is suitable for the automatic management of Android Apps in the market. The proposed system can be combined with the existing methods using DNS servers or monitoring modules to identify harmful Android apps better in different stages.

• Journal of Korean Society on Water Environment
• /
• v.35 no.2
• /
• pp.152-164
• /
• 2019

In this study, a Benthic Macroinvertebrates Family Index (BMFI) was developed using 100 indicator groups (99 families including Chironomidae with 2 phena). Families were assigned a score between 1 and 10 depending on their sensitivity to organic pollution. The BMFI was composed of the sensitivity and relative abundance of the indicator taxa. Sensitivity values of each group were generally similar to Biological Monitoring Working Party (BMWP) scores or Walley, Hawkes, Paisley, Trigg (WHPT) scores of UK, Japanese BMWP scores, and the FBI tolerance values of North America. However, sensitivity values of some taxa were significantly different from those of foreign countries, which seemed to have resulted from discrepancy in species composition, difference of taxonomic classification system, or methodological difference for estimation of sensitivity. As an annual average level, BMFI showed significant correlation with concentration of 5-day biochemical oxygen demand (BOD5) (correlation coefficient r = -0.80, n = 569 sites), total suspended solids (r = -0.68), and total phosphorus (r = -0.79). In addition, BMFI revealed strong correlation with Shannon-Weaver's species diversity (r = 0.85), Margalef's species richness (r = 0.85) and McNaughton's dominance (r = -0.84). Correlation between BMFI and water quality parameters or community indices such as species diversity did not show significant difference compared to that of species-level indices such as BMI (Benthic Macroinvertebrates Index). This means that BMFI is a more useful indicator in terms of easy identification of organisms. BMFI was used to assess the environmental status of 3,017 sites of Stream Ecosystem Survey conducted by the Korean Ministry of Environment between 2016 and 2018. As a result, about half of all sites appeared to be in good condition, and a quarter in poor condition.

• Korean Journal of Biological Psychiatry
• /
• v.27 no.1
• /
• pp.18-26
• /
• 2020

Objectives The aim was to find effective vectorization and classification models to predict a psychiatric diagnosis from text-based medical records. Methods Electronic medical records (n = 494) of present illness were collected retrospectively in inpatient admission notes with three diagnoses of major depressive disorder, type 1 bipolar disorder, and schizophrenia. Data were split into 400 training data and 94 independent validation data. Data were vectorized by two different models such as term frequency-inverse document frequency (TF-IDF) and Doc2vec. Machine learning models for classification including stochastic gradient descent, logistic regression, support vector classification, and deep learning (DL) were applied to predict three psychiatric diagnoses. Five-fold cross-validation was used to find an effective model. Metrics such as accuracy, precision, recall, and F1-score were measured for comparison between the models. Results Five-fold cross-validation in training data showed DL model with Doc2vec was the most effective model to predict the diagnosis (accuracy = 0.87, F1-score = 0.87). However, these metrics have been reduced in independent test data set with final working DL models (accuracy = 0.79, F1-score = 0.79), while the model of logistic regression and support vector machine with Doc2vec showed slightly better performance (accuracy = 0.80, F1-score = 0.80) than the DL models with Doc2vec and others with TF-IDF. Conclusions The current results suggest that the vectorization may have more impact on the performance of classification than the machine learning model. However, data set had a number of limitations including small sample size, imbalance among the category, and its generalizability. With this regard, the need for research with multi-sites and large samples is suggested to improve the machine learning models.

• Korean Journal of Construction Engineering and Management
• /
• v.22 no.6
• /
• pp.120-130
• /
• 2021

Recently, the government is reinforcing safety management at construction sites to prevent safety accidents in construction works, and the safety management plan of workers is an important situation. Meanwhile, modular construction is expected to be gradually expanded to middle and high-rise buildings, but active measures to ensure worker safety are insufficient. This study is a preliminary study of the development of a checklist for preventive worker safety management. The purpose of this study is to derive a checklist classification system for the safety management of workers in the field of modular construction by preceding studies, case analysis, and expert advisory opinions. The classification system consists of large categories of factory manufacturing, transportation, and on-site construction, and the sub-system consists of six sub-classes: foundation work, frame work, modular frame installation work, finishing work, and facility work. Among them, the sub-classification of modular frame installation work consists of 12 unit works, centering on module lifting and assembly module work, which are the main construction processes. And the classification system reflects the three main management factors and contents for defined safety management. It is expected that the research results of this study can contribute to efficient safety management at the modular construction site.

• The Journal of Society for e-Business Studies
• /
• v.24 no.2
• /
• pp.1-14
• /
• 2019

Because of the characteristics of game software, it is important to quickly identify and reflect users' needs into game software after its launch. However, most sites such as the Google Play Store, where users can download games and post reviews, provide only very limited and ambiguous classification categories for game reviews. Therefore, in this paper, we develop an automatic classification system for game reviews that categorizes reviews into categories that are clearer and more useful for game providers. The developed system converts words in reviews into vectors using word2vec, which is a representative word embedding model, and classifies reviews into the most relevant categories by measuring the similarity between those vectors and each category. Especially, in order to choose the best similarity measure that directly affects the classification performance of the system, we have compared the performance of three representative similarity measures, the Euclidean similarity, cosine similarity, and the extended Jaccard similarity, in a real environment. Furthermore, to allow a review to be classified into multiple categories, we use a threshold-based multi-category classification method. Through experiments on real reviews collected from Google Play Store, we have confirmed that the system achieved up to 95% accuracy.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.66 no.9
• /
• pp.1392-1401
• /
• 2017

In this study, the design methodology of pattern classification is introduced for avoiding faults through partial discharge occurring in the power facilities and local sites. In order to classify some partial discharge types according to the characteristics of each feature, the model is constructed by using the Radial Basis Function Neural Networks(RBFNNs) and Particle Swarm Optimization(PSO). In the input layer of the RBFNNs, the feature vector is searched and the dimension is reduced through Principal Component Analysis(PCA) and PSO. In the hidden layer, the fuzzy coefficients of the fuzzy clustering method(FCM) are tuned using PSO. Raw datasets for partial discharge are obtained through the Motor Insulation Monitoring System(MIMS) instrument using an Epoxy Mica Coupling(EMC) sensor. The preprocessed datasets for partial discharge are acquired through the Phase Resolved Partial Discharge Analysis(PRPDA) preprocessing algorithm to obtain partial discharge types such as void, corona, surface, and slot discharges. Also, when the amplitude size is considered as two types of both the maximum value and the average value in the process for extracting the preprocessed datasets, two different kinds of feature datasets are produced. In this study, the classification ratio between the proposed RBFNNs model and other classifiers is shown by using the two different kinds of feature datasets, and also we demonstrate the proposed model shows superiority from the viewpoint of classification performance.