• Journal of the Korean Society for information Management
• /
• v.34 no.4
• /
• pp.7-32
• /
• 2017

Recently, deep learning has been rapidly spreading as an innovative machine learning technique in various domains. This study explored the research trends of deep learning via modified ego centered topic citation analysis. To do that, a few seed documents were selected from among the retrieved documents with the keyword 'deep learning' from Web of Science, and the related documents were obtained through citation relations. Those papers citing seed documents were set as ego documents reflecting current research in the field of deep learning. Preliminary studies cited frequently in the ego documents were set as the citation identity documents that represents the specific themes in the field of deep learning. For ego documents which are the result of current research activities, some quantitative analysis methods including co-authorship network analysis were performed to identify major countries and research institutes. For the citation identity documents, co-citation analysis was conducted, and key literatures and key research themes were identified by investigating the citation image keywords, which are major keywords those citing the citation identity document clusters. Finally, we proposed and measured the citation growth index which reflects the growth trend of the citation influence on a specific topic, and showed the changes in the leading research themes in the field of deep learning.

• Korean Journal of Social Welfare
• /
• v.64 no.2
• /
• pp.159-183
• /
• 2012

This study intends to explore structural characteristics of social networks for female immigrants married to Korean men, and to analyze the relationship between the characteristics and types of social supports provided by their social networks and the differences between support-giving and support-receiving networks. Ego-centered network analysis is used for collecting network data on fifty-three migrant wives selected by a snowball sampling method. Results show that social support receiving and giving networks of female immigrants have similarities rather than differences, which implied that they play roles not only as support receivers, but also as support givers in their social networks. Also the study suggests that there are correlations between networks' characteristics, such as density and effective size of ego network, and types of supports. The result indicates that the less cohesive and less redundant ties female immigrants had, the more diverse and more informational and emotional supports they obtained from their social networks. Due to the sampling method and size, this study has a limitation to generalize the results for the whole population of female immigrants in Korea. However, it provides a basic understanding of female immigrants' social networks.

• Annual Conference on Human and Language Technology
• /
• 2011.10a
• /
• pp.79-82
• /
• 2011

본 논문은 다양한 주제 분야의 연구 성과물을 제공하는 학술DB에서 주제어(Keyword) 정보를 바탕으로 SNA(Social Network Analysis)기법을 적용해 검색어와 연관도가 높은 연관검색어를 제공하는 것을 그 목적으로 한다. 이를 위해 주제어들 간의 가중치(Weight)를 계산한 뒤 Ego Network 분석을 통해 검색어와 연관된 연관주제어를 추출하고 이를 기존 학술DB에서 제공한 연관검색어와 비교 정리하였다. 그리고 정리된 결과를 연관규칙 마이닝기법, 유사계수를 적용해 연관도측면에서 비교 평가하였다.

• Annals of Geriatric Medicine and Research
• /
• v.20 no.3
• /
• pp.149-159
• /
• 2016

Background: Since the mid-20th century, the ways in which social networks and older adults' health are related have been widely studied. However, few studies investigate the relationship between self-rated health and position in a complete social network of one entire Korean rural village. This study highlights use of a complete network in health studies. Methods: Using the Korean Social Life and Health Project, the population-based data of adults aged 60 or older and their spouses in one myeon in Ganghwa island (Ganghwa-gun, Incheon, Korea), Incheon, Korea (with a 95% response rate), this study built a $1,012{\times}1,012$ complete social network matrix of the village. The data were collected from 2011 to 2012, and 731 older adults were analyzed. The ordered logistic models to predict self-rated health allowed us to examine social factors from socio-demographic to individual community activities, ego-centered network characteristics, and positions in a complete network. Results: From the network data, 5 network components were identified. Even after controlling for all other factors, if a respondent belonged to a segregated component, the probability that he or she reported good health dropped substantially. Additionally, high in-degree centrality was connected to greater self-rated health. Conclusion: This finding highlights the importance of social position not only from the respondents' point of view but also from the entire village's perspective. Even if a respondent maintained a large social network, when all of those social ties belonged to a segregated group in the village, the respondent's health suffered from this segregation.

• Journal of the Korean Society for information Management
• /
• v.31 no.4
• /
• pp.161-178
• /
• 2014

Recently, many information services allow users to collaborate to produce and use information. Sharing information is also important for users who have similar taste or interest. As various channels are available for users to share their experiences and knowledge, users' data have also been accumulated within the information services. This study collected movie lists made by users of IMDB service. Co-word analysis and ego-centered network analysis were adapted to discover relevant information for users who chose a specific movie. Three factors of movies including movie title, director and genre were used to present related movie information. Movie title is an effective feature to present related movies with various aspects such as theme or characters and the popularity of directors affects on identifying related directors. Genre is not useful to find related movies due to the complexity in the topic of a movie.

• Knowledge Management Research
• /
• v.22 no.1
• /
• pp.217-242
• /
• 2021

The purpose of this study is to examine the research trends in the filed of individual knowledge hiding through keyword network analysis. As individuals intentionally hide their knowledge beyond not sharing their knowledge in organizations and the research on knowledge hiding steadily spreads, it is necessary to examine the research trends regarding knowledge hiding behaviors. For keyword network analyses, we collected 346 kinds of 578 keywords from 120 articles associated with knowledge hiding behaviors. We also transformed the keywords to 86 nodes and 667 links by data standardizing criteria and finally analyzed the keyword network among them. Moreover, this study scrutinized knowledge hiding trends by comparing the conceptual model for knowledge hiding based on literature review and the network structure based on keyword network analysis. As results, first, the network centrality degree, knowledge sharing, creativity, and performance was higher than others in Degree, Betweenness, Closeness centrality. Second, this study analyzed ego networks about psychological ownership and individual emotion theoretically associated with knowledge hiding and explored the relationship between variables through comparing with the conceptual model for knowledge hiding. Finally, the study suggested theoretical and practical implications and provided the limitations and suggestions for future research based on study findings.

• Journal of Information Science Theory and Practice
• /
• v.2 no.2
• /
• pp.16-32
• /
• 2014

As Topic Modeling has been applied in increasingly various domains, the difficulty in naming and characterizing topics also has been recognized more. This study, therefore, explores an approach of combining text mining with network analysis in a multi-step approach. The concept of synchronization was applied to re-assign the top author keywords in more than one topic category, in order to improve the visibility of the topic-author keyword network, and to increase the topical cohesion in each topic. The suggested approach was applied using 16,548 articles with 2,881 unique author keywords in construction and building engineering indexed by KSCI. As a result, it was revealed that the combined approach could improve both the visibility of the topic-author keyword map and topical cohesion in most of the detected topic categories. There should be more cases of applying the approach in various domains for generalization and advancement of the approach. Also, more sophisticated evaluation methods should also be necessary to develop the suggested approach.

• Journal of Engineering Education Research
• /
• v.20 no.5
• /
• pp.50-58
• /
• 2017

The purpose of this study is to explore the core competencies of the Fourth Industrial Revolution in the major newspaper articles of social network analysis and to examine the core competencies required by each field and target. To do this, we reviewed prior research focusing on core competency concepts and core competencies of engineering students, and analyzed 227 articles related to core competencies of the 4th Industrial Revolution, focusing on five major newspapers. Through analysis, we analyzed social network with 118 refined core competency keywords. As a result of the research, it was found that core competencies of the 4th Industrial Revolution are the degree centrality in terms 'creativity', 'problem solving ability', 'convergence ability', 'collaboration ability', 'conductivity', 'software ability', 'human literacy', 'personality' order. Also, as a result of the analysis of the ego centric network by field and target, the required core competencies of university and industry were found to be different. Through these discussions, it is necessary to restructure the core competence of engineering students in order to nurture the engineering talents necessary for the 4th Industrial Revolution.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.5
• /
• pp.2398-2415
• /
• 2017

Content-Centric Networking (CCN) is a new Internet architecture with routing and caching centered on contents. Through its receiver-driven and connectionless communication model, CCN natively supports the seamless mobility of nodes and scalable content acquisition. In-network caching is one of the core technologies in CCN, and the research of efficient caching scheme becomes increasingly attractive. To address the problem of unbalanced cache load distribution in some existing caching strategies, this paper presents a neighbor cooperation based in-network caching scheme. In this scheme, the node with the highest betweenness centrality in the content delivery path is selected as the central caching node and the area of its ego network is selected as the caching area. When the caching node has no sufficient resource, part of its cached contents will be picked out and transferred to the appropriate neighbor by comprehensively considering the factors, such as available node cache, cache replacement rate and link stability between nodes. Simulation results show that our scheme can effectively enhance the utilization of cache resources and improve cache hit rate and average access cost.

• Journal of Auto-vehicle Safety Association
• /
• v.14 no.3
• /
• pp.65-70
• /
• 2022

Lane change in urban environments is a challenge for both human-driving and automated driving due to their complexity and non-linearity. With the recent development of deep-learning, the use of the RNN network, which uses time series data, has become the mainstream in this field. Many researches using RNN show high accuracy in highway environments, but still do not for urban environments where the surrounding situation is complex and rapidly changing. Therefore, this paper proposes a lane change possibility decision network by adopting Attention layer, which is an SOTA in the field of seq2seq. By weighting each time step within a given time horizon, the context of the road situation is more human-like. A total 7D vectors of x, y distances and longitudinal relative speed of side front and rear vehicles, and longitudinal speed of ego vehicle were used as input. A total 5,614 expert data of 4,098 yield cases and 1,516 non-yield cases were used for training, and the performance of this network was tested through 1,817 data. Our network achieves 99.641% of test accuracy, which is about 4% higher than a network using only LSTM in an urban environment. Furthermore, it shows robust behavior to false-positive or true-negative objects.