Earticle

본 논문은 상대적으로 새로운 기법인 Parzen Density Estimation과 Multi-class SVM을 이용한 지능형 고장 탐색과 진단 방법을 제안하고 있다. 본 연구에서는 롤링 베어링을 대상으로 고장을 탐색하고 진단하기 위한 방법을 제안하는데 Parzen Density Estimation과 Multi-class SVM은 고장 클래스를 잘 표현할 수 있다. Parzen Density Estimation은 새로운 패턴 데이터의 거절과 알려진 데이터 패턴의 밀도의 평가에 의해 새로운 패턴을 찾아낼 수 있고, Multi-class SVM 기반의 방법은 여러 클래스의 고장을 support vector로 표현하여 고장 패턴을 찾아낼 수 있다. 본 연구에서는 실제의 다중 클래스를 가지는 롤링 베어링의 고장 데이터를 사용하여 고장 패턴을 탐색하는 과정을 보여주는데, 커널함수의 적절한 파라미터의 선택에 의한 Multi-class SVM 기반의 방법이 multi-layer perceptron이나 Parzen Density Estimation 방법보다 우수함을 입증한다.

Fault detection of rotating engine components in the aircraft engine is a challenging task that must constantly be monitored to provide aviation safety. In this paper, we propose a novel approach based on multi-layer perceptron (MLP) to detect in real time the degree of faults in a turbine engine disk due to a crack. To further improve detection accuracy while reducing computational complexity, the recursive feature elimination (RFE) is applied as a potent feature selection method. Satisfactorily, simulation results show that the proposed framework is robust to changes in operating conditions and outperforms comparative approaches.

Support vector machines (SVMs) have become one of the most popular approaches to learning from examples and have many potential applications in science and engineering. However, their applications in fault diagnosis of rotating machinery are rather limited. Most of the published papers focus on some special fault diagnoses. This study covers the overall diagnosis procedures on most of the faults experienced in rotating machinery and examines the performance of different SVMs strategies. The excellent characteristics of SVMs are demonstrated by comparing the results obtained by artificial neural networks (ANNs) using vibration signals of a fault simulator.

Internet Traffic identification is a crucial method for monitoring Internet application activities and is essential for Internet management and security. Internet traffic data typically displays imbalanced distributions. The uneven distribution of instances in each class indicates the class imbalance problem. This problem can cause a decrease in classification performance because the classifier assumes the dataset has a balanced class distribution. Internet Traffic Identification dataset is often accompanied by overlapping and noise. The hybrid approach to handling class imbalances involving data-level and ensemble-based approaches is usually chosen to overcome this problem. Data-level with oversampling using SMOTE is the choice because of its ability to synthesize new samples for minority classes. However, SMOTE-generated samples tend to be noisy and overlap with the majority of samples. This research proposes the application of a Hybrid Approach with Membership-density-based Oversampling to tackle this challenge. This research emphasizes the importance of applying membership degrees in determining samples that will group samples into safe, overlapping, and noisy areas. Then, top samples will be selected based on density ratio, stability, and score for safe and overlapping safe areas. The study findings that the proposed method effectively addresses multi-class imbalances in six Internet Traffic Identification datasets, yielding slightly improved average accuracy, and class balance accuracy results compared to other testing methods, though the difference is not statistically significant. The noise and overlapping scenes experiments demonstrate that the average accuracy obtained is superior, showing a considerable difference compared to all test methods.

Addressing imbalanced data is essential for accurate prediction. We propose a novel ensemble method of XGBoost and deep Q-learning networks (DQN) for pregnancy risk prediction. First, we train the majority class utilizing XGBoost. We subsequently utilize DQN to train the minority class into binary classifications. Finally, we use the trained models from DQN and XGBoost in ensemble learning to generate the final classification results. The XGBoost-DQN model achieves high performance with 0.9819 in precision, recall, F1-score, and accuracy, outperforming several baseline classifiers on private data from 5313 pregnant women in Indonesia and showing superior results on public datasets.

[연구목적] 본 연구는 다학년 학급의 역사적 변천과 오늘날의 모습을 바탕으로 다학년 학급의 본질적인 의미를 도출하는 것에 목적이 있다. [연구방법] 이를 위하여 로마 시대부터 중세 시대, 르네상스 시대, 산업혁명, 근대화 시대, 현 대에 이르기까지 학급의 구조를 묘사한 2차 사료를 탐색하였으며, 이를 바탕으로 학급의 구조가 변화한 과정을 기술하고, 다학년 학급의 역사 및 오늘날의 모습, 그 리고 단식학급의 발생 과정을 분석하였다. [연구결과] 다학년 학급은 학교 제도가 탄생하면서 자연스럽게 형성된 학급 형태였으며, 학생 의 연령보다 개별 학생의 수준을 고려하는 방향으로 구성되었다. 또한 다학년 학급 은 지역, 학교, 교사 수준에서 주로 형성되는 학급 구성 방식이었으며, 점차 교육 적인 의도를 추구하는 방향으로 발전하였다. [결 론] 이 연구 결과를 바탕으로 현실적으로 학급 통합이 증가하는 상황에서 다학년 학급 을 위한 국가, 지역, 학교, 교사 차원의 협력이 필요하며, 개별 학생을 고려하는 교육을 위한 대안으로써 다학년 학급을 효과적인 교육 전략으로 발전시킬 필요가 있다는 결론을 얻었다.

Purpose: This study aims to derive the essential meaning of multi-grade class based on it’s historical evolution and contemporary characteristics. Methods: To this end, secondary sources describing the structure of classes from the Roman era, the Middle Ages, the Renaissance, the Industrial Revolution, the modernization era and the modern era were explored. Based on this, the process of changes in class structures was described, and the historical development and contemporary state of multi-grade classes, as well as the emergence of mono-grade classes, were analyzed. Results: Multi-grade classes naturally emerged with the establishment of school systems and were structured to consider the individual levels of students rather than their ages. Moreover, multi-grade classes were primarily formed at the regional, school, and teacher levels and gradually evolved to pursue educational intentions. Conclusion: Based on the results of this study, it was concluded that cooperation at the national, regional, school, and teacher levels is necessary in the context of increasing class consolidation. Multi-grade classes should be developed as an effective educational strategy, providing an alternative that considers the individual needs of students.

Many studies have focused on the prediction of corporate credit rating using various data mining techniques. One of the most frequently used algorithms is support vector machines (SVM), and recently, novel techniques such as SVM+ and SVM+MTL have emerged. This paper intends to show the applicability of such new techniques to multi-classification and corporate credit rating and compare them with conventional SVM regarding prediction performance. We solve multi-class SVM+ and SVM+MTL problems by constructing several binary classifiers. Furthermore, to demonstrate the robustness and outstanding performance of SVM+MTL algorithm over other techniques, we utilized four typical multi-class processing methods in our experiments. The results show that SVM+MTL outperforms both conventional SVM and novel SVM+ in predicting corporate credit rating. This study contributes to the literature by showing the applicability of new techniques such as SVM+ and SVM+MTL and the outperformance of SVM+MTL over conventional techniques. Thus, this study enriches solving techniques for addressing multi-class problems such as corporate credit rating prediction.

토지피복도는 환경 관리와 농업 정책 수립에 필수적인 공간정보이지만 국내에서는 여전히 수작업 판독에 의존하고 있어 갱신 주기가 길고 경계 정밀도가 제한적이다. 본 연구는 고해상도 RGB 정사영상과 U-Net 기반 의미 분할 모델을 활용하여 세부 토지피복 분류 및 농경지 경계 검증의 자동화 가능성을 평가하였다. 연구 대상지는 농경 지가 우세하고 인공 구조물이 산재한 경상북도 구미시 해평면이다. 공간해상도 25 cm 정사영상을 512×512 픽셀 단 위로 분할하고 11개 클래스로 수작업 라벨링하여, 총 1,172개 타일을 학습에, 200개를 검증에 사용하였다. 검증 결과, MIoU는 0.3342, 전체 정확도는 0.8157로 나타났다. MIoU 값은 대규모 딥러닝 연구의 일반적 기준보다 낮지만 제한 된 데이터 규모와 클래스 불균형을 고려할 때 의미 있는 결과로 판단된다. 수치지도, 지적도 및 공식 토지피복도와의 시각적 비교를 통해 본 연구의 방법이 필지 단위 농경지 경계와 태양광 시설을 보다 현실적으로 추출함을 확인하였 다. 이러한 결과는 정사영상 기반 딥러닝 기법이 농촌 지역에서 토지피복도 자동 갱신과 농경지 경계 추출에 효과적 으로 활용될 수 있음을 보여준다.

Land cover maps are essential for environmental management and agricultural policy, but in Korea they are still produced mainly through manual interpretation, limiting update frequency and boundary precision. This study assesses the feasibility of automating detailed land cover classification and farmland boundary verification using high-resolution RGB orthophotos and a U-Net semantic segmentation model. The study area is Haepyeong-myeon, Gumi City, Gyeongsangbuk-do, a rural region dominated by agricultural fields with scattered artificial structures. Orthophotos at 25 cm resolution were tiled into 512 × 512 patches and manually labeled into 11 classes. In total, 1,172 tiles were used for training and 200 for validation. The trained U-Net achieved a Mean Intersection over Union(MIoU) of 0.3342 and an overall accuracy of 0.8157 on the validation set. While the mIoU is below typical benchmarks, the results remain meaningful given the limited dataset size and class imbalance. Comparisons with topographic, cadastral, and official land cover maps indicate that the proposed method delineates parcel-level farmland boundaries and solar facilities more realistically. Overall, the findings suggest that orthophoto-based deep learning can support automated land cover map updates and farmland boundary extraction in rural areas.

Support vector machines (SVMs), a machine learning technique, has been applied to not only binary classification problems such as bankruptcy prediction but also multi-class problems such as corporate credit ratings. However, in general, the performance of SVMs can be easily worse than the best alternative model to SVMs according to the selection of predictors, even though SVMs has the distinguishing feature of successfully classifying and predicting in a lot of dichotomous or multi-class problems. For overcoming the weakness of SVMs, this study has proposed an approach for selecting features for multi-class SVMs that utilize the impurity measures of classification trees. For the selection of the input features, we employed the C4.5 and CART algorithms, including the stepwise method of discriminant analysis, which is a well-known method for selecting features. We have built a multi-class SVMs model for credit rating using the above method and presented experimental results with data regarding S&P 500 companies.

본 연구는 다학년 학급의 국내외 운영현황을 살피고, 체계적으로 적용할 수 있 는 교육과정 운영 방향을 탐색하는 것을 목적으로 하였다. 이를 위하여 국내외 문헌 및 통계자료, 다학년 학급 운영이 능동적으로 실천되 고 있는 국외의 사례를 검토하여 다학년 학급의 교육과정 특징 등을 분석하였다. 분석 결과 첫째, 다학년 학급은 국내외를 막론하고 전 세계에서 쉽게 찾아볼 수 있는 학급의 형태로서, 한국의 경우 증가하는 추세이다. 둘째, 학교의 여건을 고려하여 인접학년, 격리학년, 혼합학년, 무학년 등의 학급편성이 이루어져야 하 고, 교육과정 운영을 위한 관리 구조를 마련해야 한다. 셋째, 다학년 학급의 요 구를 충족시키기 위해 단일학년을 전제로 개발된 교육과정을 재구성해야 한다. 이것은 다년간의 교육과정 범위설정, 차별화 교육과정, 준 단일학년 교육과정 (Quasi monograde), 학습자 및 자료중심 교육과정 등으로 이루어질 수 있다. 이상의 다학년 학급 교육과정에 대한 기본적인 내용들은 다양한 다학년 학급 구 성형태를 포괄할 수 있는 교육과정 개발 원리를 제시해 줄 수 있다.

The purpose of this study was to examine the domestic and international operation status of multi-grade classes, and to explore the direction of curriculum operation that can be systematically applied. To this end, we reviewed domestic and international literature and statistical data, as well as cases of multi-grade classes that are actively practiced in foreign countries, and analyzed the educational characteristics of multi-grade classes. The analysis results show that first, multi-grade classes are a type of classroom that can be easily found all over the world, regardless of domestic or foreign countries, and that the number of multi-grade classes in Korea is increasing. Second, the school's conditions should be considered, and the classroom should be organized in a variety of ways, such as adjacent grades, isolated grades, mixed grades, and non-grades, and a management structure for curriculum operation should be established. Third, the curriculum developed based on a single grade should be reorganized to meet the needs of multi-grade classes. This can be done in a variety of ways, such as setting the scope of the curriculum for multiple years, differentiated curriculum, quasi-monograde curriculum, and learner- and material-centered curriculum. The basic contents of the multi-grade class curriculum can provide principles for curriculum development that can cover a variety of multi-grade class compositions.

본 연구는 다학년 학급의 교수학습방법이 구체적으로 검토 및 연구되지 못하고 있음을 지적하고, 체계적으로 적용할 수 있는 교수학습방법을 탐색하는 것을 목 적으로 하였다. 이를 위하여 다학년 학급 운영이 능동적으로 실천되고 있는 국외의 사례를 검 토하여 다학년 학급의 교수학습 특징, 수업전략 등을 분석하였다. 분석 결과 다학년 학급의 교수학습에서는 첫째, 학생들 사이의 이질성이 더 크 기 때문에 이를 줄이거나 활용하는 수업 전략을 사용해야하고, 둘째, 구체적으로 전체 학습, 동료 교수, 집단 학습, 자기 주도 학습의 네 가지 수업 전략으로 접 근할 수 있다. 특히, 집단 학습은 집단을 구성하는 방식에 따라 능력에 따른 그 룹화(ability grouping), 혼합 능력 그룹화(mixed-ability grouping)로 구분하 며, 교사는 학급을 구성하는 학생의 특징에 따라 집단을 구성하는 기준을 결정해 야 한다. 셋째, 이상의 형태를 조합하여, 전체 학습→동료 교수→집단 학습/자기 주도 학습→전체 학습, 전체 학습→집단 학습/동료 교수→자기 주도 학습→전체 학습, 전체 학습→자기 주도 학습→집단학습→전체 학습 등으로 수업 안을 예시 하였다. 다학년 학급이 가지고 있는 고유의 장점을 살린다면 일반학급에서도 다학년 학 급 형태의 장점을 취할 수 있는 구체적인 수업방법이 마련될 수 있다. 이는 단순 히 다학년 학급의 원활한 운영을 지원하는 것을 뛰어넘어서, 건설적인 교수학습 방법 중 하나로 발전해 나가는 계기가 될 것이다.

The purpose of this study is to explore the systematic teaching and learning methods that can be applied to multi-grade classes. This is done by examining the teaching and learning characteristics and strategies of multi-grade classes in foreign countries where multi-grade class operation is actively practiced. The analysis results show that in multi-grade classes, first, students have more heterogeneity, so teaching strategies that reduce or utilize this heterogeneity should be used. Second, it can be approached in four teaching strategies: whole-class learning, peer teaching, group learning, and self-directed learning. In particular, group learning is divided into ability grouping and mixed-ability grouping depending on the way the group is formed, and the teacher should determine the criteria for forming the group according to the characteristics of the students in the class. Third, by combining the above forms, the lesson plan was exemplified as follows: whole-class learning → peer teaching → group learning/self-directed learning → whole-class learning, whole-class learning → group learning/peer teaching → self-directed learning → whole-class learning, whole-class learning → self-directed learning → group learning → whole-class learning. If the unique advantages of multi-grade classes are utilized, specific teaching methods can be prepared that can take advantage of the advantages of multi-grade classes in regular classes. This will go beyond simply supporting the smooth operation of multi-grade classes, and it will be an opportunity to develop into one of the constructive teaching and learning methods.

본 연구는 60세부터 74세까지의 독거 노인 남성의 대사증후군 복합 위험 요인을 잠재계층분석으로 분류하 고 특징을 분석했다. 국민건강영양조사에 참여한 520명을 대상으로 허리둘레, 혈압, 혈당, 중성지방, HDL 콜레스 테롤 수치를 분석했다. 그 결과 건강군(26%), 흡연 및 스트레스군(19.6%), 비만군(29.2%), 음주 및 우울군(25.2%) 의 4개 계층이 확인됐다. 로지스틱 회귀분석 결과, 흡연 및 스트레스, 비만, 음주 및 우울군은 건강군 대비 각각 4.86배, 5.11배, 2.76배 높은 대사증후군 위험을 보였다. 이는 독거 노인 남성의 특성을 고려한 맞춤형 건강 관리의 중요성을 시사한다.

This study classified multi-risk factors of metabolic syndrome in 60-74 year-old men living alone using Latent Class Analysis. Data from KNHANES on 520 men, including waist circumference, blood pressure, blood glucose, triglycerides, and HDL cholesterol, were analyzed. Four classes emerged: healthy (26%), smoking and stress (19.6%), obesity (29.2%), and alcohol and depression (25.2%). Logistic regression showed significantly higher metabolic syndrome risk in the smoking and stress (4.86 times), obesity (5.11 times), and alcohol and depression (2.76 times) groups compared to the healthy group, emphasizing the need for tailored health management for this demographic.

국내는 정보통신(IT) 기술의 발달을 기반으로 BIS, BMS, CNS 등을 통해 교통정보가 실시간적으로 제공되고 있어 유비쿼터스 기반의 첨단 통신기술이 정착되게 되면 개인의 취향에 적합한 맞춤형 (대중)교통정보가 가능할 것으로 전망된다. 대중교통 통행정보는 경로탐색알고리즘으로 구현되는데, 대중교통망은 일반 가로망과는 달리 환승과 운행시간 특성을 포함하며 통행자가 경로를 선택하는 과정에서 통행시간, 환승횟수, 환승시간과 같은 복수의 제약 또는 목적의 고려가 요구된다. 본 연구에서는 사용자에게 영향을 미칠 수 있는 요인들을 복수의 목적으로 고려하여 경로를 탐색하며 탐색된 경로에 대하여 자의적으로 선택해 나가는 기법을 제안한다 본 연구는 복합대중교통망을 대상으로 통행시간, 환승횟수, 환승시간등과 같은 통행목적에 대하여 계층별로 별도의 다목적 경로를 탐색하는 방안을 강구하여 계층별로 적용 가능한 사용자 맞춤형 정보제공체계의 기본적인 접근이론을 개발한다. 제안된 다수단, 다계층, 다목적 경로탐색알고리즘은 링크표지에 근거하여 구축되어 복합교통망에 효율적으로 구동할 수 있으며, 수도권 도시철도의 실제 대규모 교통망에 적용한 사례연구를 통하여 다양한 계층을 고려한 경로계획모형(Path Planning)으로 활용될 수 있는 가능성을 예시하였다.

Mass transit information should contribute many benefits to users. Especially transportation information technology is developing highly with information technology in Korea recently. Hereafter it is expected to provide customized transportation information to users individually with the advent of ubiquitous age in earnest. This public transportation information service can be realized by path finding algorithms in public transportation networks including travel and transfer attributes. In this research, multi objectives such as travel time, transfer time, and number of transfer and so on are constructed with the primary facts influencing users. Moreover, the method reducing user's path finding alternatives arbitrarily is proposed by selecting the best alternative which provides maximum utility to users among non dominated paths. Therefore, the ultimate goal of this study proposes a multi objective shortest paths finding algorithm which can take into account multiple user classes in a transit network with multiple travel modes. The proposed algorithm is demonstrated based on the two case studies - a small toy network and the large-scaled Seoul Metropolitan subway network.

최근 인터넷의 급격한 확장에 따른 정보의 양이 기하급수적으로 증가하고 있다. 그러나 실제 사용자에게 필요한 정보는 극히 일부분으로 사용자가 원하는 정보를 찾는데 까지는 부가적인 시간과 노력이 요구된다. 따라서 검색어로 검색된 문서에 대한 유사도 평가를 통한 계층적 유사 정보와 검색 우선순위에 대한 정보를 제공할 필요성이 있다. 이를 위해서 검색어를 구성하고 있는 키워드의 동시 발생 빈도를 고려한 검색 문서에 대한 유사도를 기반으로 문서 클러스터를 구성하고 SVM을 적용한 빅 데이터 기반 계층적 유형 분류 모델을 제안한다. 계층적 분류방법과 SVM 분류기의 결합은 문서의 계층이 기하급수적으로 늘어나는 웹 문서의 경우에 높은 성능을 얻을 수 있다. 제안된 모델은 정확하고 신속한 검색을 제공하는 정보검색시스템의 응용 모델로 활용될 수 있다.

Recently data growth rates are growing exponentially according to the rapid expansion of internet. Since users need some of all the information, they carry a heavy workload for examination and discovery of the necessary contents. Therefore information retrieval must provide hierarchical class information and the priority of examination through the evaluation of similarity on query and documents. In this paper we propose an Multi-class support vector machines model based clustering for hierarchical document categorization that make semantic search possible considering the word co-occurrence measures. A combination of hierarchical document categorization and SVM classifier gives high performance for analytical classification of web documents that increase exponentially according to extension of document hierarchy. More information retrieval systems are expected to use our proposed model in their developments and can perform a accurate and rapid information retrieval service.

For the last few decades, many studies have tried to explore and unveil venture companies’ success factors and unique features in order to identify the sources of such companies’ competitive advantages over their rivals. Such venture companies have shown tendency to give high returns for investors generally making the best use of information technology. For this reason, many venture companies are keen on attracting avid investors’ attention. Investors generally make their investment decisions by carefully examining the evalua-tion criteria of the alternatives. To them, credit rating information provided by international rating agencies, such as Standard and Poor’s, Moody’s and Fitch is crucial source as to such pivotal concerns as companies stability, growth, and risk status. But these types of information are generated only for the companies issuing corporate bonds, not venture companies. Therefore, this study proposes a method for evaluating venture businesses by presenting our recent empirical results using financial data of Korean venture companies listed on KOSDAQ in Korea exchange. In addition, this paper used multi-class SVM for the prediction of DEA-based efficiency rating for venture businesses, which was derived from our proposed method. Our approach sheds light on ways to locate efficient companies generating high level of profits. Above all, in determining effective ways to evaluate a venture firm’s efficiency, it is important to understand the major contributing factors of such efficiency. Therefore, this paper is constructed on the basis of following two ideas to classify which companies are more efficient venture companies: i) making DEA based multi-class rating for sample companies and ii) developing multi-class SVM-based efficiency prediction model for classifying all companies.First, the Data Envelopment Analysis(DEA) is a non-parametric multiple input-output efficiency technique that measures the relative efficiency of decision making units(DMUs) using a linear programming based model. It is non-parametric because it requires no assumption on the shape or parameters of the underlying production function. DEA has been already widely applied for evaluating the relative efficiency of DMUs. Recently, a number of DEA based studies have evaluated the efficiency of various types of companies, such as internet companies and venture companies. It has been also applied to corporate credit ratings. In this study we utilized DEA for sorting venture companies by efficiency based ratings.The Support Vector Machine(SVM), on the other hand, is a popular technique for solving data classification problems. In this paper, we employed SVM to classify the efficiency ratings in IT venture companies according to the results of DEA. The SVM method was first developed by Vapnik [1995]. As one of many machine learning techniques, SVM is based on a statistical theory. Thus far, the method has shown good performances especially in generalizing capacity in classification tasks, resulting in numerous applications in many areas of business. SVM is basically the algorithm that finds the maximum margin hyperplane, which is the maximum separation between classes. According to this method, support vectors are the closest to the maximum margin hyperplane. If it is impossible to classify, we can use the kernel function. In the case of nonlinear class bounda-ries, we can transform the inputs into a high-dimensional feature space. This is the original input space and is mapped into a high-dimensional dot-product space. Many studies applied SVM to the prediction of bank-ruptcy, the forecast a financial time series, and the problem of estimating credit rating. In this study we employed SVM for developing data mining-based efficiency prediction model. We used the Gaussian radial function as a kernel function of SVM. In multi-class SVM, we adopted one-against-one approach between binary classification method and two all-together methods, proposed by Weston and Watkins[1999] and Crammer and Singer[2000], respectively.In this research, we used corporate information of 154 companies listed on KOSDAQ market in Korea exchange. We obtained companies’ financial information of 2005 from the KIS(Korea Information Service, Inc.). Using this data, we made multi-class rating with DEA efficiency and built multi-class prediction model based data mining. Among three manners of multi-classification, the hit ratio of the Weston and Watkins method is the best in the test data set. In multi classification problems as efficiency ratings of venture business, it is very useful for investors to know the class with errors, one class difference, when it is difficult to find out the accurate class in the actual market. So we presented accuracy results within 1-class errors, and the Weston and Watkins method showed 85.7% accuracy in our test samples.We conclude that the DEA based multi-class approach in venture business generates more information than the binary classification problem, notwithstanding its efficiency level. We believe this model can help investors in decision making as it provides a reliably tool to evaluate venture companies in the financial domain. For the future research, we perceive the need to enhance such areas as the variable selection process, the parameter selection of kernel function, the generalization, and the sample size of multi-class.

본 연구의 목적은 혼합연령 학급 내 자연스러운 상호작용 과정에서 일어나는 의사소통 특성과 유형 을 관찰ㆍ분석하고, 대상의 연령에 따른 의사소통 유형도 분석하여 혼합연령 학급 유아들에 대한 이해 를 도모함으로써 효율적인 혼합연령 학급 운영의 기초 자료를 제공하는 데 있다. 이를 위해 만 4, 5세 혼합연령 학급에서 의사소통이 일어나는 상황을 친사회적 상황과 반사회적 상황으로 나누어 각각의 상황 속에서 나타나는 연령 간 의사소통 유형을 분석하였다. 그 결과 첫째, 만 4세에게는 더 높은 수 준의 언어모델을 제공하여 언어발달을 도모할 수 있으며, 만 4세도 상대의 언어적 수준에 따라 다양한 의사소통을 시도하고 있었다. 만 5세는 의사소통시 상대의 언어적 수준을 고려하여 언어적 행동을 조 절할 수 있음이 나타났으며, 만 4세의 반사회적 행동에도 긍정적으로 의사소통하고 있었다. 둘째, 혼합 연령 학급에서 유아들은 상호작용 대상의 연령에 따라 다양한 의사소통 유형이 나타났다. 듣는 사람이 어린 경우 사물의 특성이나 이유, 정당성 등을 설명해주는 언어와 행동을 제안하는 유형, 나이가 많은 경우에는 질문과 자신의 행동에 대한 허락을 구하는 언어유형을 많이 구사하였다.

The purpose of this study was to investigate communication characteristics and the patterns of children in the Multi-aged class. Research subjects 26 children(14 four years old, 12 five years old), which is the class for four, five years old at E Kindergarten where is located in Seoul. A research method was carried out the participatory observation and the informal interview between children and teacher. As a result of this study, first, four years old children were provided a high-level language model and five years old considered the other side of the linguistic level. Second, children in the Multi-aged class interactions depending on the age of the target appeared various communication types.

본 연구는 학생 참여형 수업과 의사소통 역량 사이의 구조적 관계에서 학생 참여와 자아개념의 매개효과를 분석하는 데 그 목적이 있다. 학생 참여형 수업의 운영에 영향을 미칠 수 있는 학급규 모(학급당 학생수)에 따라 집단을 구분하여 의사소통 역량 함양에 직, 간접적인 영향을 주는 변인들 의 차이점을 분석하여, 의사소통 역량 함양을 위한 시사점을 얻고자 하였다. 이를 위해 한국교육개 발원의 한국교육종단조사 6차년도 데이터를 이용하여 다집단 구조방정식 모형 분석을 실시하였다. 분석 결과 모든 집단에서 학생 참여형 수업방법이 의사소통 역량에 긍정적인 영향을 주었으며, 학 급규모가 작은 집단이 학급규모가 큰 집단에 비해 의사소통 역량 함양에 보다 긍정적인 영향을 받 은 것으로 나타났다. 자아개념은 모든 집단에서 의사소통 역량 함양에 긍정적인 영향을 주었으나, 학생 참여는 학급규모가 큰 집단에서만 유의했으며 그 영향이 작은 것으로 나타났다. 이와 함께 학 생 참여형 수업방법과 의사소통 역량 사이의 관계에서 자아개념의 매개효과가 있는 것으로 나타났 다. 의사소통 역량 함양을 위해 다양한 방식의 학생 참여형 수업의 도입을 고려해야 하며, 학생들의 긍정적 자아개념 형성을 위한 다차원적인 연구가 필요하다.

The purpose of this study is to analyze the mediating effect of student engagement and self-concept in the structural relationship between student participatory class and communication competency. By dividing groups according to the class size (the number of students per class) that may affect the operation of student-participatory classes, the differences in variables that directly or indirectly affect communication competency development were analyzed to obtain implications for communication competency development. To this end, a multi-group structural equation model analysis was conducted using data from the 6th year of the Korea Education Development Institute's Korean Education Longitudinal Study. As a result of the analysis, it was found that the student-participatory class method had a positive effect on communication competency in all groups, and the group with a small class size had a more positive effect on communication competency development than the group with a large class size. The self-concept had a positive effect on the development of communication skills in all groups, but student engagement was significant only in groups with a large class size, and the effect was found to be small. In addition, it was found that there was a mediating effect of self-concept in the relationship between student participatory teaching methods and communication competencies. In order to cultivate communication skills, it is necessary to consider the introduction of various student participatory teaching methods, and research on various measures to form positive self-concepts of students is necessary.

온라인 커뮤니티에서는 익명성을 기반으로 한 혐오표현이 지속적으로 발생하며, 심각한 사회문제로 이어지고 있다. 안전한 플랫폼 환경을 유지하기 위해서는 혐오표현을 탐지하는 기술이 필요하다. 그러나 언어마다 표현 방식이 달라, 다국어 데이터에서는 심각한 클래스 불균형이 발생한다. 본 연구에서는 한국어와 영어 기반의 6개 공개 데이터셋을 통합하여 9개의 레이블로 구성하였으며, 이 과정에서 극히 희소한 언어-레이블 조합이 나타났다. 이를 해결하기 위해 희소한 조합에 한정하여 로짓 조정, 소프트 타깃 학습, Group-DRO를 적용하고, 다중레이블 분류와 이진 분류를 결합한 멀티태스크 학습 구조를 도입하였다. 실험 결과, 제안 모델은 micro/macro F1에서 각각 0.79를 기록하였으며, 특히 희소 조합에서 단일 태스크 기준 대비 +0.22p와 +0.38p의 효과를 보였다. 이러한 결과는 희소 조합의 재현율 저하와 언어 간 성능 격차를 동시에 완화할 수 있음을 보여준다.

Hate speech, often fueled by the anonymity of online communities, continues to proliferate and poses a serious social problem. To maintain safe and trustworthy online platforms, effective detection of hate speech is essential. However, linguistic variations across languages lead to severe class imbalance in multilingual datasets. In this study, we integrated six publicly available datasets in Korean and English into a unified dataset with nine labels, which revealed extremely sparse language?label combinations. To address this issue, we applied logit adjustment, soft-target learning, and Group-DRO specifically to these sparse combinations, and introduced a multi-task learning framework that combines multi-label and binary classification. Experimental results show that our model achieved a micro/macro F1 of 0.79, with improvements of +0.22p and +0.38p over single-task baselines in the most sparse combinations. These findings demonstrate that the proposed approach can simultaneously mitigate recall degradation in sparse cases and reduce performance disparities across languages.