• Journal of Intelligence and Information Systems
• /
• v.25 no.4
• /
• pp.105-122
• /
• 2019

Dimensionality reduction is one of the methods to handle big data in text mining. For dimensionality reduction, we should consider the density of data, which has a significant influence on the performance of sentence classification. It requires lots of computations for data of higher dimensions. Eventually, it can cause lots of computational cost and overfitting in the model. Thus, the dimension reduction process is necessary to improve the performance of the model. Diverse methods have been proposed from only lessening the noise of data like misspelling or informal text to including semantic and syntactic information. On top of it, the expression and selection of the text features have impacts on the performance of the classifier for sentence classification, which is one of the fields of Natural Language Processing. The common goal of dimension reduction is to find latent space that is representative of raw data from observation space. Existing methods utilize various algorithms for dimensionality reduction, such as feature extraction and feature selection. In addition to these algorithms, word embeddings, learning low-dimensional vector space representations of words, that can capture semantic and syntactic information from data are also utilized. For improving performance, recent studies have suggested methods that the word dictionary is modified according to the positive and negative score of pre-defined words. The basic idea of this study is that similar words have similar vector representations. Once the feature selection algorithm selects the words that are not important, we thought the words that are similar to the selected words also have no impacts on sentence classification. This study proposes two ways to achieve more accurate classification that conduct selective word elimination under specific regulations and construct word embedding based on Word2Vec embedding. To select words having low importance from the text, we use information gain algorithm to measure the importance and cosine similarity to search for similar words. First, we eliminate words that have comparatively low information gain values from the raw text and form word embedding. Second, we select words additionally that are similar to the words that have a low level of information gain values and make word embedding. In the end, these filtered text and word embedding apply to the deep learning models; Convolutional Neural Network and Attention-Based Bidirectional LSTM. This study uses customer reviews on Kindle in Amazon.com, IMDB, and Yelp as datasets, and classify each data using the deep learning models. The reviews got more than five helpful votes, and the ratio of helpful votes was over 70% classified as helpful reviews. Also, Yelp only shows the number of helpful votes. We extracted 100,000 reviews which got more than five helpful votes using a random sampling method among 750,000 reviews. The minimal preprocessing was executed to each dataset, such as removing numbers and special characters from text data. To evaluate the proposed methods, we compared the performances of Word2Vec and GloVe word embeddings, which used all the words. We showed that one of the proposed methods is better than the embeddings with all the words. By removing unimportant words, we can get better performance. However, if we removed too many words, it showed that the performance was lowered. For future research, it is required to consider diverse ways of preprocessing and the in-depth analysis for the co-occurrence of words to measure similarity values among words. Also, we only applied the proposed method with Word2Vec. Other embedding methods such as GloVe, fastText, ELMo can be applied with the proposed methods, and it is possible to identify the possible combinations between word embedding methods and elimination methods.

• Journal of Korean Philosophical Society
• /
• no.116
• /
• pp.251-281
• /
• 2017

The Parmenidean tradition of logos which previous researches fail to fully appreciate has three dimensions of reality-knowledge-discourse. Parmenides is not just an ontologist, as the traditional view emphasizes, but also an epistemologist, as the revisionist view begins to emphasize, and, at the same time, a meta-discourser, as those two established views fail to embrace. In order to reach the third view which fully grasps such a dynamic and integrated feature of Parmenides, we should closely pay attention to the organic interconnectedness of three discourse parts of truth-doxa-proem, especially the significance of proem and meta-discourse. In the Eleatic tradition of discourse, the figure who clearly appreciated and further developed such an authentic feature of Parmenides' discourse is not, as one might easily expect, one of the second-generation Eleatics, but Gorgias who has commonly been positioned at the opposite side of Eleatism. This paper investigates how he actually both innovated and succeeded the Parmenidean tradition of logos; especially, it characterizes his discourse as an antilogy(antilogia) from within the tradition: as a 'devil' advocate' who complemented and completed Parmenidean persuasion by positing the Parmenidean tradition of logos as an arena of a huge intellectual discipline and cultivation, offering himself as a sparring partner to it, and bringing up an antilogy. In the process of this antilogy he performed in his rhetorical speeches such as the Encomium of Helen and the Defense of Palamedes he experimented and examined a possibility of persuasion operating independently from truth, which, however, is not merely sacrificing truth in favor of persuasiveness and probability (to eikos) as Plato criticized mainly focussing on his 'philosophical' writing On not-being. Rather, it was an 'opposition for opposition's sake' and serious play which purported to provide balance and flexibility to contemporary intellectual society which had too much inclined towards truth and knowledge and become stiff and to put weight on the opposite side of mainstream. It is wholly our eranos (i.e. our share of contribution) to summon and examine such sophistic tradition for the sake of the task of our times, not for the sake of Plato's task, that we should build up a healthy culture of discourse where we can share serious play.

• Implantology
• /
• v.22 no.4
• /
• pp.242-254
• /
• 2018

The aim of the current study was to evaluate the results of horizontal guided bone regeneration (GBR) with xenograf t (deproteinized bovine bone mineral, DBBM), allograf t (irradiated allogenic cancellous bone and marrow), titanium membrane, resorbable collagen membrane, and advanced platelet-rich fibrin (A-PRF) in the anterior maxilla. The titanium membrane was used in this study has a three-dimensional (3D) shape that can cover ridge defects. Case 1. A 32-year-old female patient presented with discomfort due to mobility and pus discharge on tooth #11. Three months after extracting tooth #11, diagnostic software (R2 GATE diagnostic software, Megagen, Daegu, Korea) was used to establish the treatment plan for implant placement. At the first stage of implant surgery, GBR for horizontal augmentation was performed with DBBM ($Bio-Oss^{(R)}$, Geistlich, Wolhusen, Switzerland), irradiated allogenic cancellous bone and marrow (ICB $cancellous^{(R)}$, Rocky Mountain Tissue Bank, Denver, USA), 3D-titanium membrane ($i-Gen^{(R)}$, Megagen, Daegu, Korea), resorbable collagen membrane (Collagen $membrane^{(R)}$, Genoss, Suwon, Korea), and A-PRF because there was approximately 4 mm labial dehiscence after implant placement. Five months after placing the implant, the second stage of implant surgery was performed, and healing abutment was connected after removal of the 3D-titanium membrane. Five months after the second stage of implant surgery was done, the final prosthesis was then delivered. Case 2. A 35-year-old female patient presented with discomfort due to pain and mobility of implant #21. Removal of implant #21 fixture was planned simultaneously with placement of the new implant fixture. At the first stage of implant surgery, GBR for horizontal augmentation was performed with DBBM ($Bio-Oss^{(R)}$), irradiated allogenic cancellous bone and marrow (ICB $cancellous^{(R)}$), 3D-titanium membrane ($i-Gen^{(R)}$), resorbable collagen membrane (Ossix $plus^{(R)}$, Datum, Telrad, Israel), and A-PRF because there was approximately 7 mm labial dehiscence after implant placement. At the second stage of implant surgery six months after implant placement, healing abutment was connected after removing the 3D-titanium membrane. Nine months after the second stage of implant surgery was done, the final prosthesis was then delivered. In these two clinical cases, wound healing of the operation sites was uneventful. All implants were clinically stable without inflammation or additional bone loss, and there was no discomfort to the patient. With the non-resorbable titanium membrane, the ability of bone formation in the space was stably maintained in three dimensions, and A-PRF might influence soft tissue healing. This limited study suggests that aesthetic results can be achieved with GBR using 3D-titanium membrane and A-PRF in the anterior maxilla. However, long-term follow-up evaluation should be performed.

• Journal of Nuclear Fuel Cycle and Waste Technology(JNFCWT)
• /
• v.17 no.3
• /
• pp.339-346
• /
• 2019

Based on the $8^{th}$ Basic Plan for Electric Power Demand and Supply, an estimation has been made for inventories and characteristics of spent fuel (SF) to be generated from existing and planned nuclear power plants. The characteristics under consideration in this study are dimensions, fuel array, $^{235}U$ enrichment, discharge burnup, and cooling time for each fuel assembly. These are essentially needed for designing a disposal facility for SFs. It appears that the anticipated quantity by the end of 2082 is about 62,500 assemblies for PWR SFs. The inventories of Westinghouse-type and Korean-type SFs were revealed to be 60% and 40%, respectively as of the end of 2018. The proportion of SFs with initial $^{235}U$ enrichment below 4.5 weight percent (wt%) was shown to be approximately 90% in total as of the end of 2018. As of 2077, more than 97% of SFs generated from Westinghouse-type nuclear reactors were shown to have cooling time of over 50 years. As of 2125, more than 98% of SFs generated from Korean-type nuclear reactors were shown to have cooling time of over 45 years. Based on these results, for the efficient design of a disposal system, it is reasonable to adopt two types of reference spent fuel. SF of KSFA with $^{235}U$ enrichment of 4.5 wt%, discharge burnup of 55 GWd/tU, and cooling time of 50 years was determined as reference fuel for Westinghouse-type SFs; SF of PLUS7 with $^{235}U$ enrichment of 4.5 wt%, discharge burnup of 55 GWd/tU, and cooling time of 45 years was determined as reference fuel for Korean-type SFs.

• Journal of The Korean Association For Science Education
• /
• v.41 no.3
• /
• pp.237-250
• /
• 2021

Teachers play a key role in designing a students' learning experience. Teachers are asked to interpret the context in which they are located and to adjust their practice to fit circumstantial needs based on their teacher agency. In this study, we explore the emergence of teacher agency in distance learning caused by COVID-19 and we analyze factors shaping the teacher agency. For this purpose, we interviewed six secondary science teachers who practiced distance learning in 2020. Semi-constructed interviews and their artifacts were collected and analyzed. This study shows that teacher agency is captured when they respond to circumstantial change and modify their practice to achieve their professional purpose or adjust their practice in space for maneuvering or keep their practice consistent. This study also analyzes the factors that affect the emergence of teacher agency in two dimensions. One is individual and the other is contextual. In the individual dimension, educational values shaped by his/her experiences and short/long-term goals for the future support the emergence of teacher agency. In the contextual dimension, there are collaborative and flexible culture shared by the community, co-operation within the teacher community, and material support. On the other hand, in the individual dimension, the teachers' sense of their role, and no reflection for own practice constrain the emergence of teacher agency, and in the contextual dimension, performativity discourse and strong requirement without guidance constrain the emergence of teacher agency. We suggest an effective lens for establishing a strategy that support teachers' professional practice and the emergence of teacher agency.

• Journal of Korean Society of Forest Science
• /
• v.110 no.2
• /
• pp.198-209
• /
• 2021

This study evaluated the effects of the dimensional characteristics of containers on the nitrogen status of Quercus serrata, Fraxinus rhynchophylla, and Zelkova serrata in the container nursery stage. Seedlings were grown using 16 container types [four growing densities (100, 144, 196, and 256 seedlings/m²) × four cavity volumes (220, 300, 380, and 460 cm³/cavity)]. Two-way ANOVA was performed to test the differences in nitrogen concentration and seedling content among container types. Additionally, we performed multiple regression analyses to correlate container dimensions and nitrogen content. Container types had a strong influence on nitrogen concentration and the content of the seedling species, with a significant interaction effect between growing density and cavity volume. Cavity volumes were positively correlated with the nitrogen content of the three seedling species, whereas growing density negatively affected those of F. rhynchophylla. Further, nutrient vector analysis revealed that the seedling nutrient loading capacities of the three species, such as efficiency and accumulation, were altered because of the different fertilization effects by container types. The optimal ranges of container dimension by each tree species, obtained multiple regression analysis with nitrogen content, were found to be approximately 180-210 seedlings/m² and 410-460 cm³/cavity for Q. serrata, 100-120 seedlings/m² and 350-420 cm³/cavity for F. rhynchophylla, and 190-220 seedlings/m² and 380-430 cm³/cavity for Z. serrata. This study suggests that an adequate type of container will improve seedling quality with higher nutrient loading capacity production in nursery stages and increase seedling growth in plantation stages.

• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.175-197
• /
• 2021

Recently, with the development of deep learning technology, research on unstructured data analysis is being actively conducted, and it is showing remarkable results in various fields such as classification, summary, and generation. Among various text analysis fields, text classification is the most widely used technology in academia and industry. Text classification includes binary class classification with one label among two classes, multi-class classification with one label among several classes, and multi-label classification with multiple labels among several classes. In particular, multi-label classification requires a different training method from binary class classification and multi-class classification because of the characteristic of having multiple labels. In addition, since the number of labels to be predicted increases as the number of labels and classes increases, there is a limitation in that performance improvement is difficult due to an increase in prediction difficulty. To overcome these limitations, (i) compressing the initially given high-dimensional label space into a low-dimensional latent label space, (ii) after performing training to predict the compressed label, (iii) restoring the predicted label to the high-dimensional original label space, research on label embedding is being actively conducted. Typical label embedding techniques include Principal Label Space Transformation (PLST), Multi-Label Classification via Boolean Matrix Decomposition (MLC-BMaD), and Bayesian Multi-Label Compressed Sensing (BML-CS). However, since these techniques consider only the linear relationship between labels or compress the labels by random transformation, it is difficult to understand the non-linear relationship between labels, so there is a limitation in that it is not possible to create a latent label space sufficiently containing the information of the original label. Recently, there have been increasing attempts to improve performance by applying deep learning technology to label embedding. Label embedding using an autoencoder, a deep learning model that is effective for data compression and restoration, is representative. However, the traditional autoencoder-based label embedding has a limitation in that a large amount of information loss occurs when compressing a high-dimensional label space having a myriad of classes into a low-dimensional latent label space. This can be found in the gradient loss problem that occurs in the backpropagation process of learning. To solve this problem, skip connection was devised, and by adding the input of the layer to the output to prevent gradient loss during backpropagation, efficient learning is possible even when the layer is deep. Skip connection is mainly used for image feature extraction in convolutional neural networks, but studies using skip connection in autoencoder or label embedding process are still lacking. Therefore, in this study, we propose an autoencoder-based label embedding methodology in which skip connections are added to each of the encoder and decoder to form a low-dimensional latent label space that reflects the information of the high-dimensional label space well. In addition, the proposed methodology was applied to actual paper keywords to derive the high-dimensional keyword label space and the low-dimensional latent label space. Using this, we conducted an experiment to predict the compressed keyword vector existing in the latent label space from the paper abstract and to evaluate the multi-label classification by restoring the predicted keyword vector back to the original label space. As a result, the accuracy, precision, recall, and F1 score used as performance indicators showed far superior performance in multi-label classification based on the proposed methodology compared to traditional multi-label classification methods. This can be seen that the low-dimensional latent label space derived through the proposed methodology well reflected the information of the high-dimensional label space, which ultimately led to the improvement of the performance of the multi-label classification itself. In addition, the utility of the proposed methodology was identified by comparing the performance of the proposed methodology according to the domain characteristics and the number of dimensions of the latent label space.

• Korea Science and Art Forum
• /
• v.30
• /
• pp.151-161
• /
• 2017

In the fields of science and engineering, multidisciplinary research is common, and researchers with a diverse range of expertise collaborate to achieve common goals. As the 4th industrial revolution gains currency in society, there is growing demand on talented personnel both with technical knowledge and skills and with communicative skills. That is, future engineers are expected to possess competence in social and artistic skills in addition to specialized knowledge and skills in engineering. In this paper we introduce an emerging field of 3D bioprinting as an exemplary case of interdisciplinary research. We have chosen the case to demonstrate the possibility of cultivating engineers with π-shaped expertise. Building on the concept of T-shaped talent, we define π-shaped expertise as having both technical skills in more than one specialized field and interpersonal/communicative skills. Wtih references to such concepts as trading zones and interactional expertise, we suggest that π-shaped expertise can be cultivated via the creation of multi-level trading zones. Trading zones are referred to as the physical, conceptual, or metaphorical spaces in which experts with different world views trade ideas, objects, and the like. Interactional expertise is cultivated, as interactions between researches are under way, with growing understanding of each other's expertise. Under the support of the university and the government, two researchers with expertise in printing technology and life sciences cooperate to develop a 3D bioprinting system. And the primary investigator of the research laboratory under study has aimed to create multiple dimensions of trading zones where researchers with different educational and cultural backgrounds can exchange ideas and interact with each other. As 3D bioprinting has taken shape, we have found that a new form of expertise, namely π-shaped expertise is formed.

• Asia-Pacific Journal of Business Venturing and Entrepreneurship
• /
• v.18 no.3
• /
• pp.61-79
• /
• 2023

This study is a case study of overseas banana farms as a global agricultural startup that has hardly been attempted so far in terms of paradigm shift in the industry, beyond regional limitations. It was researched for the purpose of revealing the success factors of 'global agricultural startup' in terms of business process, entrepreneurship, and management dimensions learned through direct participation and observation at the local level. In order to study global agricultural startups, this study also conducted a comparative analysis of global startups (global startups) and global agricultural startups(global agricultural startups). In fact, the analysis consists of 'definition', 'components', and 'success factors', and we want to confirm the difference between the two concepts that can be distinguished. The case analysis tried to maximize the advantages of 'participatory action research' by directly observing and experiencing banana farms. In the case of banana farm cases, by dividing them into preparation process for farm development and farm development and management process, various variables considered in farm management were explained through the whole process of farm management. Through the process of overcoming and responding to specific failure cases, we tried to secure the reliability and validity of the research, and the case studies related to entrepreneurship, management, and organization analyzed by applying them by subdividing them into theoretical areas belonging to components and management that were theorized in existing preceding studies. This study is almost the first study on the process of creating a local entry business by directly moving the head office overseas rather than entering overseas agriculture as a subsidiary, joint venture or overseas corporation. In particular, it is a unique case that corresponds to agriculture in terms of region(Africa), scale(startup), and industry that have not been introduced so far as a global agricultural startup. In terms of entrepreneurship, it also concretely exemplified how entrepreneurship components such as innovativeness, risk-taking propensity, proactiveness, vision sharing, social contribution, leadership, etc., which have not been attempted so far in agricultural cases, are manifested and effective. The management and cultural aspects also went beyond the argument that only cultural aspects are important in overseas business, and also confirmed individual failure cases and their responses in recruitment, job, wage, retirement, development, organizational structure management, etc. As a result, there is significance and implications of this study in that it provides theoretical confirmation as well as practical and responsive basis for 'entrepreneurship', 'farming management', and 'management' aspects in overseas agricultural startup business operation.

• Journal of Practical Engineering Education
• /
• v.16 no.5_spc
• /
• pp.747-756
• /
• 2024

The purpose of this study was to discover the desirable ESG direction of public institutions and obstacles to promoting ESG management as perceived by members within public institutions. To this end, three research questions were set and FGI was conducted on eight groups of public institution members. As a result of the FGI data analysis, 9 subfactors and 43 meaning units in 3 dimensions corresponding to the answers to the 3 research questions were derived. The first dimension, public institution members' awareness of ESG, consisted of three subfactors: ESG was recognized as an environmental protection trend related to the climate change crisis, and as a marketing tool used by companies to enhance their image. In addition, it was recognized as a newly included public institution management evaluation index. The second dimension, obstacles to the promotion of ESG in public institutions, appeared to be four subfactors: a government-dependent management system strongly influenced by the government, a rigid internal communication system in a top-down manner, the possibility of lack of sincerity in promoting ESG management, limitations of the internal human resource management system, etc. The third dimension, the desirable ESG direction of public institutions, was found to be two subfactors, including priority promotion of activities that meet the unique purpose of public institutions and ESG activities that can grow together with the region as a member of the local community. This study is significant in that, unlike existing studies that discovered positive antecedent factors that affect a company's ESG management performance, it identified factors that impede performance achievement from the perspective of the members who drive ESG.