• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.12
• /
• pp.1595-1603
• /
• 2020

Online defamation incidents such as Internet news comments on portal sites, SNS, and community sites are increasing in recent years. Bias and hate expressions threaten online service users in various forms, such as invasion of privacy and personal attacks, and defamation issues. In the past few years, academia and industry have been approaching in various ways to solve this problem The purpose of this study is to build a dataset and experiment with deep learning classification modeling for detecting various bias expressions as well as hate expressions. The dataset was annotated 7 labels that 10 personnel cross-checked. In this study, each of the 7 classes in a dataset of about 137,111 Korean internet news comments is binary classified and analyzed through deep learning techniques. The Proposed technique used in this study is multi-channel CNN model with attention. As a result of the experiment, the weighted average f1 score was 70.32% of performance.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.21 no.3
• /
• pp.99-105
• /
• 2021

In addition to the OTT video production service represented by Nexflix and YouTube, a personalized recommendation system for content with artificial intelligence has become common. YouTube's personalized recommendation service system consists of two neural networks, one neural network consisting of a recommendation candidate generation model and the other consisting of a ranking network. Netflix's video recommendation system consists of two data classification systems, divided into content-based filtering and collaborative filtering. As the online platform-led content production is activated by the Corona Pandemic, the field of virtual influencers using artificial intelligence is emerging. Virtual influencers are produced with GAN (Generative Adversarial Networks) artificial intelligence, and are unsupervised learning algorithms in which two opposing systems compete with each other. This study also researched the possibility of developing AI platform based on individual recommendation and virtual influencer (metabus) as a core content of OTT in the future.

• The Journal of Information Systems
• /
• v.30 no.4
• /
• pp.277-301
• /
• 2021

Purpose The purpose of this study is to analyze the sentiment responses of the general public to non-face-to-face work using text mining methodology. As the number of non-face-to-face complaints is increasing over time, it is difficult to review and analyze in traditional methods such as surveys, and there is a limit to reflect real-time issues. Approach This study has proposed a method of the research model, first by collecting and cleansing the data related to non-face-to-face work among tweets posted on Twitter. Second, topics and keywords are extracted from tweets using LDA(Latent Dirichlet Allocation), a topic modeling technique, and changes for each section are analyzed through DTM(Dynamic Topic Modeling). Third, the complaints of non-face-to-face work are analyzed through the classification of positive and negative polarity in the COVID-19 section. Findings As a result of analyzing 1.54 million tweets related to non-face-to-face work, the number of IDs using non-face-to-face work-related words increased 7.2 times and the number of tweets increased 4.8 times after COVID-19. The top frequently used words related to non-face-to-face work appeared in the order of remote jobs, cybersecurity, technical jobs, productivity, and software. The words that have increased after the COVID-19 were concerned about lockdown and dismissal, and business transformation and also mentioned as to secure business continuity and virtual workplace. New Normal was newly mentioned as a new standard. Negative opinions found to be increased in the early stages of COVID-19 from 34% to 43%, and then stabilized again to 36% through non-face-to-face work sentiment analysis. The complaints were, policies such as strengthening cybersecurity, activating communication to improve work productivity, and diversifying work spaces.

• Journal of the Korea Convergence Society
• /
• v.13 no.3
• /
• pp.227-237
• /
• 2022

With the growing interest in convergence, there have been various attempts to measure convergence, but the definition of convergence is ambiguous and consensus on appropriate indicators has not been reached, so measurement of convergence is still at a rudimentary stage. In this study, using the KSIC-IPC linkage table developed by the Korean Intellectual Property Office to analyze the correlation and impact of patents, industry, economy, and population, we propose a new evaluation model that can evaluate industry convergence from patent data. In addition, it was verified whether the industry convergence derived from this properly reflects the corporate convergence characteristics. As a result of classifying the convergence of 39,740 patents owned by global major automobile companies, and evaluating the degree of convergence of each company, it was confirmed that the industry convergence derived using the KSIC-IPC linkage table better reflects the corporate convergence characteristics than the technology convergence classified by IPC co-classification. Therefore, the industry convergence data of automotive sector derived from the new industry convergence evaluation model using the KSIC-IPC linkage table is expected to be widely used for future convergence research.

• Journal of Korea Trade
• /
• v.23 no.5
• /
• pp.45-65
• /
• 2019

Purpose - As the well-known Structure-Conduct-Performance paradigm implies, risk structure of a corporation may affect its risk management activity and the activity may in turn determine its performance. Depending on its goal, Foreign Direct Investment (FDI) can shape its risk structure, risk management and its performance. Under this assumption, we investigate the relationship between the goals of FDI and risk management for the first time in academics. Design/methodology - This empirical research uses a survey of 279 current Korean enterprises' FDIs in China with the recently developed business risk quadrant model. Companies are classified into either a market- or an efficiency-seeking group, to identify how each group perceives and manages risks, and values the performance of risk management. Also, we apply integrated risk management method that multinational corporations have introduced in China, then verify the mediating effect between risk factors and performance. Findings - Our research shows the FDIs can expose themselves to differing risk structure although risk management activities simply represent the level of empowerment given to local management by headquarter due to limit of sample size despite diversity of risk and risk management tools. To sum, market seekers are found to have more strategic risk (revenue related risk) than efficiency seekers with financial risk (cost related risk). The market seekers can manage their risk by empowering their local organisation while the efficiency does the opposite ways. The risk management appears to be successful in general. Originality/value - Previous studies on small and medium enterprises' FDIs to China have concentrated on the analysis of entry determinants, withdrawal factors and individual risk management. Meanwhile, this research establishes enterprise-wide risk factors faced by the companies that advance into China, according to the method of the classification by ERM and verifies if they could synthetically improve performance through risk corresponding measures.

• Herbal Formula Science
• /
• v.30 no.4
• /
• pp.281-291
• /
• 2022

Objectives : In this study, acute oral toxicity test of pomace Schisandra chinensis extracts was conducted in order to up-cycling to a high value-added industry using by-products discarded in the production process of Schisandra chinensis products and active ingredients such as dibenzocyclooctadiene lignans in Schisandra chinensis. Methods : Pomace Schisandra chinensis extracts were orally administered to SD-rats(female, n=3) without a control group according to the 'OECD guidelines'. After, mortality and clinical signs were observed, and the deceased animals were subjected to an autopsy. In addition, acute oral toxicity test was sequentially performed in step I (300 mg/kg), step II(300 mg/kg), step III(2,000 mg/kg), and step IV(2,000 mg/kg) according to the mortality. Results : There were no abnormalities caused by pomace Schisandra chinensis extracts in step I and step II. However, one animal each died in step III and step IV. In addition, clinical signs(salivation, decrease in food intake, prone position, decrease of locomotor activity, loss of locomotor activity, convulsion, hypothermia, lacrimation, staining around mouth, soiled perineal region, reddish urine, chromaturia, decrease of fecal volume, lying on side, blackish stool, no stool, compound-colored stool, refusal to feed, excitement, hypersensitivity, rigidity, dorsal position, etc.) were observed. But, no clinical signs were observed from 5th day, and experiment animals recovered completely. Conclusions : As a result of this study, pomace Schisandra chinensis extracts may exhibit acute toxicity at concentrations of 2,000 to 5,000 mg/kg, and the GHS classification was designated as 'Category 5'.

• Journal of Korean Society of Water and Wastewater
• /
• v.36 no.3
• /
• pp.149-165
• /
• 2022

The water consumption data of individual consumers must be analyzed and forecast to establish an effective water demand management plan. A k-mean cluster model that can monitor water use characteristics based on hourly water consumption data measured using automated meter reading devices and demographic factors is developed in this study. In addition, the quantification model that can estimate the daily water consumption is developed. K-mean cluster analysis based on the four clusters shows that the average silhouette coefficient is 0.63, also the silhouette coefficients of each cluster exceed 0.60, thereby verifying the high reliability of the cluster analysis. Furthermore, the clusters are clearly classified based on water usage and water usage patterns. The correlation coefficients of four quantification models for estimating water consumption exceed 0.74, confirming that the models can accurately simulate the investigated demographic data. The statistical significance of the models is considered reasonable, hence, they are applicable to the actual field. Because the use of automated smart water meters has become increasingly popular in recent year, water consumption has been metered remotely in many areas. The proposed methodology and the results obtained in this study are expected to facilitate improvements in the usability of smart water meters in the future.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.22 no.4
• /
• pp.97-113
• /
• 2023

Swift and appropriate responses in emergency situations like objects falling on the road can bring convenience to road users and effectively reduces secondary traffic accidents. In Korea, current intelligent transportation system (ITS)-based detection systems for emergency road situations mainly rely on loop detectors and CCTV cameras, which only capture road data within detection range of the equipment. Therefore, a new detection method is needed to identify emergency situations in spatially shaded areas that existing ITS detection systems cannot reach. In this study, we propose a ResNet-based algorithm that detects and classifies emergency situations from vehicle camera footage. We collected front-view driving videos recorded on Korean highways, labeling each video by defining the type of emergency, and training the proposed algorithm with the data.

• Restorative Dentistry and Endodontics
• /
• v.48 no.2
• /
• pp.16.1-16.11
• /
• 2023

Objectives: The aim of this study was to evaluate the influence of peak kilovoltage (kVp) and a metal artifact reduction (MAR) tool on image quality and the diagnosis of vertical root fracture (VRF) in cone-beam computed tomography (CBCT). Materials and Methods: Twenty single-rooted human teeth filled with an intracanal metal post were divided into 2 groups: control (n = 10) and VRF (n = 10). Each tooth was placed into the socket of a dry mandible, and CBCT scans were acquired using a Picasso Trio varying the kVp (70, 80, 90, or 99), and the use of MAR (with or without). The examinations were assessed by 5 examiners for the diagnosis of VRF using a 5-point scale. A subjective evaluation of the expression of artifacts was done by comparing random axial images of the studied protocols. The results of the diagnoses were analyzed using 2-way analysis of variance and the Tukey post hoc test, the subjective evaluations were compared using the Friedman test, and intra-examiner reproducibility was evaluated using the weighted kappa test (α = 5%). Results: The kVp and MAR did not influence the diagnosis of VRF (p > 0.05). According to the subjective classification, the 99 kVp protocol with MAR demonstrated the least expression of artifacts, while the 70 kVp protocol without MAR led to the most artifacts. Conclusions: Protocols with higher kVp combined with MAR improved the image quality of CBCT examinations. However, those factors did not lead to an improvement in the diagnosis of VRF.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.23 no.4
• /
• pp.115-119
• /
• 2023

The natural language processing models used in current artificial intelligence are huge, causing various difficulties in processing and analyzing data in real time. In order to solve these difficulties, we proposed a method to improve the efficiency of processing by using less memory and checked the performance of the proposed model. The technique applied in this paper to evaluate the performance of the proposed model is to divide the large corpus by adjusting the number of attention heads and embedding size of the BERT[1] model to be small, and the results are calculated by averaging the output values of each forward. In this process, a random offset was assigned to the sentences at every epoch to provide diversity in the input data. The model was then fine-tuned for classification. We found that the split processing model was about 12% less accurate than the unsplit model, but the number of parameters in the model was reduced by 56%.