KJC Korea
Journal Central

Recently, since most of the research on correcting speech recognition errors is based on English, there is not enough research on Korean speech recognition. Compared to English speech recognition, however, Korean speech recognition has many errors due to the linguistic characteristics of Korean language, such as Korean Fortis and Korean Liaison, thus research on Korean speech recognition is needed. Furthermore, earlier works primarily focused on editorial distance algorithms and syllable restoration rules, making it difficult to correct the error types of Korean Fortis and Korean Liaison. In this paper, we propose a context-sensitive post-processing model of speech recognition using a LSTM-based sequence-to-sequence model and Bahdanau attention mechanism to correct Korean speech recognition errors caused by the pronunciation. Experiments showed that by using the model, the speech recognition performance was improved from 64% to 77% for Fortis, 74% to 90% for Liaison, and from 69% to 84% for average recognition than before. Based on the results, it seems possible to apply the proposed model to real-world applications based on speech recognition.

Humans are social animals that have gained information or social interaction through dialogue. In conversation, the mood of the word can change depending on the sensibility of one person to another. Relationships between characters in films are essential for understanding stories and lines between characters, but methods to extract this information from films have not been investigated. Therefore, we need a model that automatically analyzes the relationship aspects in the movie. In this paper, we propose a method to analyze the relationship between characters in the movie by utilizing deep learning techniques to measure the emotion of each character pair. The proposed method first extracts main characters from the movie script and finds the dialogue between the main characters. Then, to analyze the relationship between the main characters, it performs a sentiment analysis, weights them according to the positions of the metabolites in the entire time intervals and gathers their scores. Experimental results with real data sets demonstrate that the proposed scheme is able to effectively measure the emotional relationship between the main characters.

The performance of deep learning-based malware detection and classification models depends largely on how to construct a feature set to be applied to training. In this paper, we propose an approach to select the optimal feature set to maximize detection performance for CNN-based Android malware detection. The features to be included in the feature set were selected through the Chi-Square test algorithm, which is widely used for feature selection in machine learning and deep learning. To validate the proposed approach, the CNN model was trained using 36 characteristics selected for the CICANDMAL2017 dataset and then the malware detection performance was measured. As a result, 99.99% of Accuracy was achieved in binary classification and 98.55% in multiclass classification.

Local search has been used to solve various combinatorial optimization problems. One of the most important factors in local search is the method of generating a neighbor solution. In this paper, we propose neighbor generation strategies of local search for permutation-based combinatorial optimization, and compare the performance of each strategies targeting the traveling salesman problem. In this paper, we propose a total of 10 neighbor generation strategies. Basically, we propose 4 new strategies such as Rotation in addition to the 4 strategies such as Swap which have been widely used in the past. In addition, there are Combined1 and Combined2, which are made by combining basic neighbor generation strategies. The experiment was performed by applying the basic local search, but changing only the neighbor generation strategy. As a result of the experiment, it was confirmed that the performance difference is large according to the neighbor generation strategy, and also confirmed that the performance of Combined2 is the best. In addition, it was confirmed that Combined2 shows better performance than the existing local search methods.

Using the acoustic features of speech, important social and linguistic information about the speaker can be obtained, and one of the key features is the dialect. A speaker's use of a dialect is a major barrier to interaction with a computer. Dialects can be distinguished at various levels such as phonemes, syllables, words, phrases, and sentences, but it is difficult to distinguish dialects by identifying them one by one. Therefore, in this paper, we propose a lightweight Korean dialect classification model using only MFCC among the features of speech data. We study the optimal method to utilize MFCC features through Korean conversational voice data, and compare the classification performance of five Korean dialects in Gyeonggi/Seoul, Gangwon, Chungcheong, Jeolla, and Gyeongsang in eight machine learning and deep learning classification models. The performance of most classification models was improved by normalizing the MFCC, and the accuracy was improved by 1.07% and F1-score by 2.04% compared to the best performance of the classification model before normalizing the MFCC.

In this paper, we propose a 3D design method of the vacuum forming method of the front part to improve the lightness and production efficiency of agricultural electric vehicles. For agricultural electric vehicles, lightness and production efficiency are more important than the strength of materials for collision protection. In this paper, we propose a vacuum forming design method that can replace complex machining processes such as laser machining, bending, and painting. The main purpose of this research is to improve product stability, productivity and convenience through 3D design of the front part and development of vacuum forming mold technology. Research procedure follows the 3D modeling of the front part using CATIA, finite element analysis for the structural stability using ABAQUS, manufacturing prototype for the investigation of the dimensions using 3D scanner and actual driving test under agricultural electric vehicle usage environment. The results verifies the proposed 3D design method of the vacuum forming method and are expected to be widely used by agricultural workers through the simplification of the production process of agricultural electric vehicles.

In this paper, we propose a system that can perform dumbbell exercise by recognizing the weight of dumbbells without wearing and device. With the development of virtual reality technnology, many studies are being conducted to simulate the pysical feedback of the real world in the virtual world. Accurate motion recognition is important to the elderly for rehabilitation exercises. They cannot lift heavy dumbbells. For rehabilitation exercise, correct body movement according to an appropriate weight must be performed. We use a machine learning algorithm for the accuracy of motion data input in real time. As an experiment, we was test three types of bicep, double, shoulder exercise and verified accuracy of exercise. In addition, we made a virtual gym game to actually apply these exercise in virtual reality.

Nowadays, a lot of educational TV programs have focused on the program providers’ perspective, namely through goods-dominant (G-D) logic. However, the growing development of current technologies such as IPTV with AR and VR has opened opportunities to go beyond a provider-centered perspective and to add the co-creation of value, since value can be created through the interaction between learners and the TV programs themselves. With this in mind, learner-centered TV programs, which are based on service-dominant (S-D) logic, should be offered through educational TV channels. This conceptual paper summarizes, analyzes, and synthesizes the present status of IPTV in education and suggests a new instructional approach for utilizing TV in future education based on value co-creation through learner participation. Three principles (service-dominant logic as an educational service design principle, goal-based scenario for instructional design, IPTV as state-of-the-art technology innovation) are suggested for designing educational IPTV programs.

In our society, the current smart devices have been commercialized and more time than most people are in contact with smart devices than the time in contact with the PC, it shows the same trend in terms of the game industry. Thus in the study, it was to try to make a game by utilizing the characteristics of only the smart devices in mobile environments among them by using the gyro sensor implementing the movements of characters. In particular, undergraduate student made various attempts to implement a single game using the sensor used in smart devices. In this paper, we have planned the adventure 3D running game that allows users to easily fun to play. Our goal is to implement one stage. We have discussed that you have designed and implemented.

In this paper, presents an efficient education method for training specialized edutainment SW education instructors and drone realistic media creators, not just training to acquire certificates through drone manipulation training. To this end, the NCS-based curriculum was derived. The developed curriculum includes the edutainment drone curriculum and the realistic media creator curriculum. Among them, core responsibilities were defined for the drone control curriculum and core tasks, knowledge, and attitudes were described for each. After that, a detailed curriculum for drone control was derived. In the realistic media creator curriculum, pilot education was conducted to actually produce advertisement videos to foster experts who can work directly in the industrial field. Finally, through holding an online conference in a metaverse environment, a virtual conference was operated to share and discuss media videos produced by trainees. After the end of education, the efficiency of this curriculum was proved through education satisfaction analysis for 46 education graduates. This paper presented a method to achieve internalization of SW education in non-face-to-face online education that our society must solve after post-COVID-19. In addition, an efficient educational method in a realistic media environment was suggested by showing a realistic media creator training curriculum, pilot programs, and metaverse conference management cases.

In this paper, the possibility of a UE (User Equipment) incarceration attack using RRCRejecet and RRCRelease in 5G SNPN (Standalone Non-Public Network) is analyzed based on the 3GPP standard document. First, the cell selection and reselection procedures of the UE are analyzed, and then the processing process of the false base station and the UE before and after transmission of RRCReject and RRCRelease is analyzed. As a result of the analysis, it is possible that the false base station that transmits a strong signal causes the victim UE to establish an RRC connection to the false base station itself. In addition, if the false base station transmits an RRCReject message without integrity protection in response to the victim UE's attempt to establish an RRC connection, it is determined that the victim UE can continue to stay in the RRC connection attempt process. On the other hand, it is determined that it is impossible to incarcerate the victim UE by inducing an attempt to establish an RRC connection to another false base station using RRCRelease.

In this paper, we analyze Shin's proposed dynamic ID-based user authentication scheme for TMIS(Telecare Medicine Information System), and Shin's authentication scheme is vulnerable to smart card loss attacks, allowing attackers to acquire user IDs, which enables user impersonation attack. In 2019, Shin's proposed authentication scheme attempted to generate a strong random number using ECC, claiming that it is safe to lose a smart card because it is impossible to calculate random number r'i due to the difficulty of the ECC algorithm without knowing random number ri. However, after analyzing Shin's authentication scheme in this paper, the use of transmission messages and smart cards makes it easy to calculate random numbers r'i, which also enables attackers to generate session keys. In addition, Shin's authentication scheme were analyzed to have significantly greater overhead than other authentication scheme, including vulnerabilities to safety analysis, the lack of a way to pass the server's ID to users, and the lack of biometric characteristics with slightly different templates.

In this paper, we propose the educational model for a convergence minor by fusing culinary arts with robot technology to develop coding ability for the students in the culinary arts major which is not originally related to software field. It is meaningful that the educational model follows the trend along the development of the fourth industrial revolution technology and has the function to make the students who are not in software major grow as software experts. However there are difficulties in designing the convergence minor because the culinary arts major is distant to the robot technology in the view of technology. To overcome this difficulty the convergence minor is designed to attract the interest for the students in culinary arts major by construct educational subjects systematically such as cooking, dessert making, barista working, autonomous serving and so on based on robots. Also the practices in which various robots are utilized are included in the convergence minor to develop actual coding ability. By comparison to the other models of convergence minors, the proposed model shows enhanced educational effects in 20% than the others.

In this paper, we propose CANVAS (Creative ANalytics enVironment And System), an analytics system of the National Research Data Platform (DataON). CANVAS is a personalized analytics cloud service for researchers who need computing resources and tools for research data analysis. CANVAS is designed in consideration of scalability based on micro-services architecture and was built on top of open-source software such as eGovernment Standard framework (Spring framework), Kubernetes, and JupyterLab. The built system provides personalized analytics environments to multiple users, enabling high-speed and large-capacity analysis by utilizing high-performance cloud infrastructure (CPU/GPU). More specifically, modeling and processing data is possible in JupyterLab or GUI workflow environment. Since CANVAS shares data with DataON, the research data registered by users or downloaded data can be directly processed in the CANVAS. As a result, CANVAS enhances the convenience of data analysis for users in DataON and contributes to the sharing and utilization of research data.

This study was intended to develop a prototype of new Hanbok Cheollik one-piece based on design extracted through a preference survey on new Hanbok Cheollik one-piece. Through this, it was intended to provide information on patterns that are fundamental to the production of new Hanbok Cheollik one piece. The pattern of the experiment was produced by modifying D pattern, which was selected as excellent in the comparison of commercial Cheollik one-piece patterns. The SPSS 26.0 program was used to analyze the appearance evaluation of patterns. As a result of the 1st evaluation of appearance and garment pressure, the shoulder, sleeve length, skirt length and sleeve width required modification, reducing the sleeve length by 8.0cm and the sleeve width by 1.0cm in total. The length of the skirt was reduced by 5.0cm and the shoulder end point was reduced by 0.5cm on both sides to modify the shoulder width. As a result of the 2nd evaluation, the waist area, sleeve width, and skirt wrinkles were required to be corrected, reducing 2.0 cm waist width and 1.0 cm sleeve width and removing wrinkles on the front center and side area. The final pattern was highly appreciated. In future studies, it is thought that research should be conducted through study of Cheollik one-piece pattern according to material and age and the actual wearing experiment according to fabric and age.

In this paper, by using the cattle feed intake, rumination, and in heat monitoring technology, RFI (Residual Feed Intake) monitoring and wearable devices and PCs for predicting abnormalities in budding target web and smart A monitoring system using a phone application was designed and implemented. With the development of this system, the farmer is expected to increase economic efficiency. By analyzing the feed intake, it is possible to identify the difference between the recommended feed amount based on the cow's weight and the feed amount consumed by the cow, and it is expected that early detection of metabolic disorders (abnormality of metabolism) is possible. Farmers using the results of this thesis can distinguish the cows with the most efficient performance, and the 6-axis motion sensor signals input from the wearable device attached to the cow's skin (neck) and the microphone attached to the wearable device. It is possible to measure the cow's rumination and feed intake through the sound of the cow's throat. In the future, improvements will be made to measure additional vital signs such as heart rate and respiration.

If one wants to file a lawsuit against the administrative office, he or she should decide whether to file a civil lawsuit or an administrative lawsuit. The type of lawsuit must be determined to determine which court to file the lawsuit with. Korea seems to have a clear distinction between administrative and judicial legal relationships, but it is not easy to distinguish between public and judicial cases unless the public and judicial discrimination are maintained. The practice or precedent of litigation is always difficult to distinguish because the litigation is based on the discrimination of whether the litigation belongs to a legal relationship in public law or judicial law. I believe that if the administrative litigation law establishes a provision related to the designation of a duty and stipulates that "if a litigation case is questioned whether it is an administrative or civil lawsuit, the Supreme Court-related court shall designate the competent court at the request of the parties," the lower court will be guaranteed the right to swift a trial, and the legal representatives will be freed from the exhaustive agony.

Recently, studies using deep learning to analyze a large amount of text are being actively conducted. In particular, a pre-trained language model that applies the learning results of a large amount of text to the analysis of a specific domain text is attracting attention. Among various pre-trained language models, BERT(Bidirectional Encoder Representations from Transformers)-based model is the most widely used. Recently, research to improve the performance of analysis is being conducted through further pre-training using BERT's MLM(Masked Language Model). However, the traditional MLM has difficulties in clearly understands the meaning of sentences containing new words such as newly coined words. Therefore, in this study, we newly propose NTM(Newly coined words Target Masking), which performs masking only on new words. As a result of analyzing about 700,000 movie reviews of portal 'N' by applying the proposed methodology, it was confirmed that the proposed NTM showed superior performance in terms of accuracy of sensitivity analysis compared to the existing random masking.

After the Ferry Sewol accident, public interest in marine safety has increased. However, as the marine leisure tourism population increases, the number of casualties caused by marine accidents is increasing, so marine safety education is urgently needed. Since facilities related to marine safety education in Korea are geographically biased to the west, regional imbalances in education are significant. Therefore, this study suggested solutions to the problems of developing educational contents and securing budgets and professionals by using idle facilities of the Korea Institute of Maritime and Fisheries Technology(KIMFT) located in Busan as a maritime safety education center. In addition, as a result of estimating demand using the gravity model, it was estimated that the demand would range from 150,000 to 130,000 per year. This study sufficiently proved social policy validity for policy suggestions using existing idle sites as maritime safety education centers based on objective verification methods and is expected to contribute substantially to policy promotion in the future.

Culture and art can be said to be an important medium that enhances people's creativity and satisfies our desire for cultural experience as well as emotional cultivation. These arts and culture are spreading all over the world through Hallyu, etc. to this day, and it is naturally positioned as an industry that creates the highest added value. However, rapid industrialization and development of industrialization also provided many negative causes, such as damage to nature due to the decline of factories and equipment industries, pollution and damage to the surrounding environment due to the use of fossil fuels. Therefore, in this study, we investigated the factors necessary for improving the image of the local community and maximizing tourism classes by actively utilizing the idle space abandoned due to industrialization and industrialization as a complex space for culture and art. As a result, empirical analysis was conducted through a total of six factors excluding typology through factor analysis. The results showed that reliability, certainty, and empathy had a positive effect, but responsiveness did not. In addition, the economic effect was found to have a positive effect on customer satisfaction.

In this paper, we propose the deep learning-based neural network model to predict bunker price. In the shipping industry, since fuel oil accounts for the largest portion of ship operation costs and its price is highly volatile, so companies can secure market competitiveness by making fuel oil purchasing decisions based on rational and scientific method. In this paper, short-term predictive analysis of HSFO 380CST in Singapore is conducted by using three recurrent neural network models like RNN, LSTM, and GRU. As a result, first, the forecasting performance of RNN models is better than LSTM and GRUs using long-term memory, and thus the predictive contribution of long-term information is low. Second, since the predictive performance of recurrent neural network models is superior to the previous studies using econometric models, it is confirmed that the recurrent neural network models should consider nonlinear properties of bunker price. The result of this paper will be helpful to improve the decision quality of bunker purchasing.

With the development of Internet technology, because traditional recommendation algorithms cannot learn the in-depth characteristics of users or items, this paper proposed a recommendation algorithm based on the AMITI(attention mechanism and improved TF-IDF) to solve this problem. By introducing the two-layer attention mechanism into the CNN, the feature extraction ability of the CNN is improved, and different preference weights are assigned to item features, recommendations that are more in line with user preferences are achieved. When recommending items to target users, the scoring data and item type data are combined with TF-IDF to complete the grouping of the recommendation results. In this paper, the experimental results on the MovieLens-1M data set show that the AMITI algorithm improves the accuracy of recommendation to a certain extent and enhances the orderliness and selectivity of presentation methods.

In this paper, the effect of artificial intelligence convergence education program that provides STEAM education using machine learning platform on elementary school students' STEAM literacy and learning flow was analyzed. A homogeneous group of 44 elementary school 6th graders was divided into an experimental group and a control group. The control group received 10 lessons of general subject convergence class, and the experimental group received 10 lessons of STEAM-based artificial intelligence convergence education using Machine learning for Kids. To develop the artificial intelligence convergence education program, the goals, achievement standards, and content elements of the 2015 revised curriculum to select subjects and class contents is analyzed. As a result of the STEAM literacy test and the learning flow test, there was a significant difference between the experimental group and the control group. In particular, it can be confirmed that the coding environment in which the artificial intelligence function is expanded has a positive effect on learners' learning flow and STEAM literacy. Among the sub-elements of convergence talent literacy, significant differences were found in the areas of personal competence such as convergence and creativity. Among the sub-elements of learning flow, significant differences were found in the areas such as harmony of challenge and ability, clear goals, focus on tasks, and self-purposed experiences. If further expanded research is conducted in the future, it will be a basic research for more effective education for the future.