• Korean Journal of Heritage: History & Science
• /
• v.47 no.2
• /
• pp.4-19
• /
• 2014

In Joseon dynasty, Gongsin-Gyoseo(Royal Edicts rewarding meritorious vassals) was issued for the reward of the contribution to the specific event of the nation. The family decided as vassals succeeded this Gongsin-Gyoseo as the family treasure to later generations. When one happened to lose this caused by an unavoidable situation such as war in this process, there were a few cases of reissuing of Gongsin-gyoseo by the nation. The confirmative reissued Gongsin-gyoseo among Gongsin-gyoseo that descended down by now are three pieces which is Park dongnyang Hoseong Gongsin-Gyoseo, Gu goeng Jeongsa Gongsin-Gyoseo, and Park jung Jeongsa Gongsin-Gyoseo. This research considered the form and mounting about these three pieces. It was indicated that the reissued Gongsin-Gyoseo was produced following the same form with the original Gongsin-Gyoseo, that is pyeongchul(moving to next lines) and daedu (writing one or two letters ahead compared to the first letter) and so on. But, it was indicated that the change of the position of the vassals in the reissued time was reflected compared to position of the original Gongsin-Gyoseo. In case of Jeonsa Gongsin-Gyoseo, it was indicated that it was 'Wonjongdaewang-whi' since the original Gongsin-Gyoseo was 'Jeongwongun-bu' and the reissued time he became the father of King Injo. Also, it was confirmed that the case of deprivation because of conspiracy such as 'Kim jajeom, Kim ryeon, Sim kiwon, Sim kisung' was produced as the original document and deleted with black ink stick. It was newly confirmed while the original Gongsin-Gyoseo was produced by a designated writer from the nation, the reissuance was written by a writer from the family of the vassals. Since the mounting of Gongsin-Gyoseo could be changed according to the favor and technique of the craftsman participated in the practical production such as Baezeopjang and economical situation of the country supplying the material, the mounting of these three Gongsin-Gyoseo should be different from the original mounting, especially because of the loss of the original by the Manchu war of 1636. The comparison result of Gongsin-Gyoseo produced in the same period with the original issued one in the record of the related Uigwe (a collection of documents from the Joseon Dynasty), the reissued one seems to be larger in the form or ratio of the mounting compared to the first issued one. First of all, the width of Byunah was expanded as twice bigger, center and both side Hoejang also was bigger as over 2cm, and the below Hoejang was expanded as 10cm, and the ratio of the upper Hoejang and the below Hoejang was wider as 1.5 times and the reissuance was 1:1 ratio. The bisect of upper shaft in Park dongnyang Hoseong Gongsin-Gyoseo is assumed as the form of an equilateral triangle, not a half-moon shape of the present, and Gu goeng and Park jung Jeongsa Gongsin-Gyoseo will be the form whose bisection form is same but the size is smaller. Chuksu is confirmed that the size is not changed significantly. Osaekdahoe can be assumed that the width was smaller compared to the first issued one. The 3 pieces of the reissued versions provide the clue of the verification for the form of the mounting of Bosa Gongsin-Gyoseo in the same production period. In the situation that the mounting of the production time was not confirmed among the currently descended Bosa Gongsin-Gyoseo, they can be very important materials.

• The Korean Journal of Archival Studies
• /
• no.32
• /
• pp.127-176
• /
• 2012

Metadata is well recognized as one of the foundational factors in archiving and long-term preservation of digital resources. There are several metadata standards for records management, archives and preservation, e.g. ISAD(G), EAD, AGRkMs, PREMIS, and OAIS. Consideration is important in selecting appropriate metadata standards in order to design metadata schema that meet the requirements of a particular archival system. Interoperability of metadata with other systems should be considered in schema design. In our previous research, we have presented a feature analysis of metadata standards by identifying the primary resource lifecycle stages where each standard is applied. We have clarified that any single metadata standard cannot cover the whole records lifecycle for archiving and preservation. Through this feature analysis, we analyzed the features of metadata in the whole records lifecycle, and we clarified the relationships between the metadata standards and the stages of the lifecycle. In the previous study, more detailed analysis was left for future study. This paper proposes to analyze the metadata schemas from the viewpoint of tasks performed in the lifecycle. Metadata schemas are primarily defined to describe properties of a resource in accordance with the purposes of description, e.g. finding aids, records management, preservation and so forth. In other words, the metadata standards are resource- and purpose-centric, and the resource lifecycle is not explicitly reflected in the standards. There are no systematic methods for mapping between different metadata standards in accordance with the lifecycle. This paper proposes a method for mapping between metadata standards based on the tasks contained in the resource lifecycle. We first propose a Task Model to clarify tasks applied to resources in each stage of the lifecycle. This model is created as a task-centric model to identify features of metadata standards and to create mappings among elements of those standards. It is important to categorize the elements in order to limit the semantic scope of mapping among elements and decrease the number of combinations of elements for mapping. This paper proposes to use 5W1H (Who, What, Why, When, Where, How) model to categorize the elements. 5W1H categories are generally used for describing events, e.g. news articles. As performing a task on a resource causes an event and metadata elements are used in the event, we consider that the 5W1H categories are adequate to categorize the elements. By using these categories, we determine the features of every element of metadata standards which are AGLS, AGRkMS, PREMIS, EAD, OAIS and an attribute set extracted from DPC decision flow. Then, we perform the element mapping between the standards, and find the relationships between the standards. In this study, we defined a set of terms for each of 5W1H categories, which typically appear in the definition of an element, and used those terms to categorize the elements. For example, if the definition of an element includes the terms such as person and organization that mean a subject which contribute to create, modify a resource the element is categorized into the Who category. A single element can be categorized into one or more 5W1H categories. Thus, we categorized every element of the metadata standards using the 5W1H model, and then, we carried out mapping among the elements in each category. We conclude that the Task Model provides a new viewpoint for metadata schemas and is useful to help us understand the features of metadata standards for records management and archives. The 5W1H model, which is defined based on the Task Model, provides us a core set of categories to semantically classify metadata elements from the viewpoint of an event caused by a task.

• The Journal of Korean Philosophical History
• /
• no.35
• /
• pp.309-334
• /
• 2012

This study attempts to study in what form Folk Taoism in the late of $Chos{\breve{o}}n$ Dynasty has existed and discuss the contents and characteristics of ideological aspects forming the foundation of private Taoism. While Guan Yu Belief(關帝信仰) in the late of $Chos{\breve{o}}n$ Dynasty is a folk belief focusing on Guan Yu, Seoneumjeulgyo(善陰?敎) and Musangdan(無相壇) are religious groups with organization. In case of Seoneumjeulgyo(善陰?敎), 'Seoneumjeul' contains perspective of Tian(天觀) of Confucianism but the ascetic practice method is to practice by reciting the name of the Buddha and the targets of a belief are Gwanje, Munchang, Buwoo. This shows the unified phenomenon of Confucianism, Buddhism, Taoism of Folk Taoism in the late of $Chos{\breve{o}}n$ Dynasty. Guan Yu Belief started at the national level led by the royal family of $Chos{\breve{o}}n$ after Japanese Invasion of Korea in 1592 was firmly settled in non-official circles. Guan Yu in the late of $Chos{\breve{o}}n$ Dynasty is expressed as the incarnation of loyalty and filial piety as well as God controlling life, death and fate. As this divine power and empowerment were spreading as scriptures among people, Guan Yu Belief was settled as a target to defeat the evil and invoke a blessing. Seoneumjeulgyo is the religious group that imitated 'Paekryunsa(白蓮社)' of Ming Qing time of China. Seoneumjeulgyo emphasized 'sympathy' with God through chanting. And it expressed writing written in the state of religious ecstasy as 'Binan(飛鸞).' Binan is also called as revelation and means to be revealed from heaven in the state united with God. Seoneumjeulgyo pursued the state united with God through a recitation of a spell and made scriptures written in the state united with God as its central doctrine. Musangdan published and spread Nanseo(鸞書,Book written by the revelation from God) and Seonso(善書) while worshipping Sam Sung Je Kun(三聖帝君). The scriptures of Folk Taoismin the late of $Chos{\breve{o}}n$ Dynasty can be roughly divided into Nanseo(鸞書) and Seonso(善書). Nanseo is a book written by the revelation from God and Seonso is a book to the standards of good deeds and encourage a person to do them such as Taishangganyingbian(太上感應篇) and Gonghwagyuk(功過格). The characteristics of Folk Taoism in the late of $Chos{\breve{o}}n$ Dynasty are as follows. First, a shrine of Guan Yu built for political reasons played a central role of Folk Taoism in the late of $Chos{\breve{o}}n$ Dynasty. Second, specific private Taoist groups such as Temple $Myory{\breve{o}}nsa$ and Musangdan appeared in the late of $Chos{\breve{o}}n$ Dynasty. These are Nandan Taoism(鸞壇道敎) that pursued the unity of God through 'sympathy' with God. Third, private Taoism of $Chos{\breve{o}}n$ was influenced by the unity of Confucianism, Buddhism, Taoism with private Taoism in the Qing Dynasty of China and religious organization form etc. Fourth, the Folk Taoism scriptures of $Chos{\breve{o}}n$ are divided into Nanseo and Seonso and Nanseo directly made in $Chos{\breve{o}}n$ is expected to be the key to reveal the characteristics of Folk Taoism.

• (The)Study of the Eastern Classic
• /
• no.56
• /
• pp.35-73
• /
• 2014

This paper examined the poem world of Simjae(深齋) Cho Geung-seop(曺兢燮: 1873-1933) in the turning point of the modern era, focused on his Hwaunsi (和韻詩: Poems written by using the rhymes of other poets' poems). In his poems, there are lots of Hwaunsi on the poems of Tu Fu(杜甫) and Su Shi(蘇軾), especially. This makes him regarded as a medieval poet, engaged in Chinese poem creation in the most traditional method in the turbulent period. Looking at the Hawunsi(和韻詩) alone, Simjae's creative life became the starting point of turnaround at around 40 years old. Before the age of 40, the poets in the Tang Dynasty and Song Dynasty and Ming Dynasty and Qing Dynasty and Korean figures like Lee Hwang(李滉), as well as Tu Fu and Su Shi were the subjects of his Hwanunsi. After the age of 40, some examples of writing poems using the rhymes of other poets' poems, especially Korean figures related to regions, are often found, reducing Hwaunsi on Tu Fu and Su Shi. Simjae called Tu Fu the integration of poets, talking about the integrity of poetic talent and his being highly proficient in mood and view. As reflecting such an awareness, the themes and moods and views are demonstrated diversely in Simjae's Hwaunsi. Although, he did not reveal his thinking about the poems of Su Shi, he seemed to love Su Shi's poems to some degree. The closeness to the original poems, the poems of Tu Fu are relatively higher than those of Su Shi. Roughly speaking, Simjae tried to find his own individuality, intending to follow Tu Fu, but, he seemed to attempt to reveal his intention using Su Shi's poems, rather than trying to imitate. To carefully examine, Simjae wrote Hwaunsi, but he did not just imitate, but revealed the aesthetics of comparison and difference. In many cases, he made new meanings by implanting his intentions in the poems, while sharing the opportunity of creation, rather than bringing the theme and mood and view as they are. The Hwaunsi on Su Shi's poems reveal the closeness to the original poems relatively less. This can be the trace of an effort to make his own theme and individuality, not being dominated by the Hwaun(和韻: using the rhymes of other poets' poems) entirely, as he used the creative method having many restrictions. However, it is noted that the Hwaunsi on Tu Fu's poems was not written much, after the age of 40. Is this the reason why he realized literary reality that he could not cope with anymore with only his effort within the Hwaunsi? For example, he wrote four poems by borrowing Su Shi's Okjungsi(獄中詩: poem written in jail) rhymes and also wrote Gujung Japje(拘中雜題), in 1919, while he was detained. In these poems, his complex contemplation and emotion, not restricted by any poet's rhymes, are revealed diversely. Simjae's Hwaunsi testifies the reality, in which Chinese poetry's habitus existed and the impressive existence mode at the turning point of the modern era. Although, the creation of Hwaunsi reflects his disposition of liking the old things, it is judged that his psychology, resisting modern characters' change, affected to some degree in the hidden side. In this regard, Simaje's Hwaunsi encounters limitation on its own, however, it has significance in that some hidden facts were revealed in the modern Chinese poetry history, which was captured with attention under the name of novelty, eccentricity and modernity.

• Journal of Intelligence and Information Systems
• /
• v.26 no.4
• /
• pp.27-65
• /
• 2020

Many companies on information and communication technology make public their own developed AI technology, for example, Google's TensorFlow, Facebook's PyTorch, Microsoft's CNTK. By releasing deep learning open source software to the public, the relationship with the developer community and the artificial intelligence (AI) ecosystem can be strengthened, and users can perform experiment, implementation and improvement of it. Accordingly, the field of machine learning is growing rapidly, and developers are using and reproducing various learning algorithms in each field. Although various analysis of open source software has been made, there is a lack of studies to help develop or use deep learning open source software in the industry. This study thus attempts to derive a strategy for adopting the framework through case studies of a deep learning open source framework. Based on the technology-organization-environment (TOE) framework and literature review related to the adoption of open source software, we employed the case study framework that includes technological factors as perceived relative advantage, perceived compatibility, perceived complexity, and perceived trialability, organizational factors as management support and knowledge & expertise, and environmental factors as availability of technology skills and services, and platform long term viability. We conducted a case study analysis of three companies' adoption cases (two cases of success and one case of failure) and revealed that seven out of eight TOE factors and several factors regarding company, team and resource are significant for the adoption of deep learning open source framework. By organizing the case study analysis results, we provided five important success factors for adopting deep learning framework: the knowledge and expertise of developers in the team, hardware (GPU) environment, data enterprise cooperation system, deep learning framework platform, deep learning framework work tool service. In order for an organization to successfully adopt a deep learning open source framework, at the stage of using the framework, first, the hardware (GPU) environment for AI R&D group must support the knowledge and expertise of the developers in the team. Second, it is necessary to support the use of deep learning frameworks by research developers through collecting and managing data inside and outside the company with a data enterprise cooperation system. Third, deep learning research expertise must be supplemented through cooperation with researchers from academic institutions such as universities and research institutes. Satisfying three procedures in the stage of using the deep learning framework, companies will increase the number of deep learning research developers, the ability to use the deep learning framework, and the support of GPU resource. In the proliferation stage of the deep learning framework, fourth, a company makes the deep learning framework platform that improves the research efficiency and effectiveness of the developers, for example, the optimization of the hardware (GPU) environment automatically. Fifth, the deep learning framework tool service team complements the developers' expertise through sharing the information of the external deep learning open source framework community to the in-house community and activating developer retraining and seminars. To implement the identified five success factors, a step-by-step enterprise procedure for adoption of the deep learning framework was proposed: defining the project problem, confirming whether the deep learning methodology is the right method, confirming whether the deep learning framework is the right tool, using the deep learning framework by the enterprise, spreading the framework of the enterprise. The first three steps (i.e. defining the project problem, confirming whether the deep learning methodology is the right method, and confirming whether the deep learning framework is the right tool) are pre-considerations to adopt a deep learning open source framework. After the three pre-considerations steps are clear, next two steps (i.e. using the deep learning framework by the enterprise and spreading the framework of the enterprise) can be processed. In the fourth step, the knowledge and expertise of developers in the team are important in addition to hardware (GPU) environment and data enterprise cooperation system. In final step, five important factors are realized for a successful adoption of the deep learning open source framework. This study provides strategic implications for companies adopting or using deep learning framework according to the needs of each industry and business.

• Maritime Security
• /
• v.5 no.1
• /
• pp.1-53
• /
• 2022

One manifestation of the contemporary era of renewed great power competition has been the deepening relationship between China and Russia. Their strengthening military ties, notwithstanding their lack of a formal defense alliance, have been especially striking. Since China and Russia deploy two of the world's most powerful navies, their growing maritime cooperation has been one of the most significant international security developments of recent years. The Sino-Russian naval exercises, involving varying platforms and locations, have built on years of high-level personnel exchanges, large Russian weapons sales to China, the Sino-Russia Treaty of Friendship, and other forms of cooperation. Though the joint Sino-Russian naval drills began soon after Beijing and Moscow ended their Cold War confrontation, these exercises have become much more important during the last decade, essentially becoming a core pillar of their expanding defense partnership. China and Russia now conduct more naval exercises in more places and with more types of weapons systems than ever before. In the future, Chinese and Russian maritime drills will likely encompass new locations, capabilities, and partners-including possibly the Arctic, hypersonic delivery systems, and novel African, Asian, and Middle East partners-as well as continue such recent innovations as conducting joint naval patrols and combined arms maritime drills. China and Russia pursue several objectives through their bilateral naval cooperation. The Treaty of Good-Neighborliness and Friendly Cooperation Between the People's Republic of China and the Russian Federation lacks a mutual defense clause, but does provide for consultations about common threats. The naval exercises, which rehearse non-traditional along with traditional missions (e.g., counter-piracy and humanitarian relief as well as with high-end warfighting), provide a means to enhance their response to such mutual challenges through coordinated military activities. Though the exercises may not realize substantial interoperability gains regarding combat capabilities, the drills do highlight to foreign audiences the Sino-Russian capacity to project coordinated naval power globally. This messaging is important given the reliance of China and Russia on the world's oceans for trade and the two countries' maritime territorial disputes with other countries. The exercises can also improve their national military capabilities as well as help them learn more about the tactics, techniques, and procedures of each other. The rising Chinese Navy especially benefits from working with the Russian armed forces, which have more experience conducting maritime missions, particularly in combat operations involving multiple combat arms, than the People's Liberation Army (PLA). On the negative side, these exercises, by enhancing their combat capabilities, may make Chinese and Russian policymakers more willing to employ military force or run escalatory risks in confrontations with other states. All these impacts are amplified in Northeast Asia, where the Chinese and Russian navies conduct most of their joint exercises. Northeast Asia has become an area of intensifying maritime confrontations involving China and Russia against the United States and Japan, with South Korea situated uneasily between them. The growing ties between the Chinese and Russian navies have complicated South Korean-U.S. military planning, diverted resources from concentrating against North Korea, and worsened the regional security environment. Naval planners in the United States, South Korea, and Japan will increasingly need to consider scenarios involving both the Chinese and Russian navies. For example, South Korean and U.S. policymakers need to prepare for situations in which coordinated Chinese and Russian military aggression overtaxes the Pentagon, obligating the South Korean Navy to rapidly backfill for any U.S.-allied security gaps that arise on the Korean Peninsula. Potentially reinforcing Chinese and Russian naval support to North Korea in a maritime confrontation with South Korea and its allies would present another serious challenge. Building on the commitment of Japan and South Korea to strengthen security ties, future exercises involving Japan, South Korea, and the United States should expand to consider these potential contingencies.

• Economic and Environmental Geology
• /
• v.56 no.5
• /
• pp.603-618
• /
• 2023

This study focused on evaluating the suitability of the WRK (waste repository Korea) bentonite as a buffer material in the SNF (spent nuclear fuel) repository. The U (uranium) adsorption/desorption characteristics and the adsorption mechanisms of the WRK bentonite were presented through various analyses, adsorption/desorption experiments, and kinetic adsorption modeling at various pH conditions. Mineralogical and structural analyses supported that the major mineral of the WRK bentonite is the Ca-montmorillonite having the great possibility for the U adsorption. From results of the U adsorption/desorption experiments (intial U concentration: 1 mg/L) for the WRK bentonite, despite the low ratio of the WRK bentonite/U (2 g/L), high U adsorption efficiency (>74%) and low U desorption rate (<14%) were acquired at pH 5, 6, 10, and 11 in solution, supporting that the WRK bentonite can be used as the buffer material preventing the U migration in the SNF repository. Relatively low U adsorption efficiency (<45%) for the WRK bentonite was acquired at pH 3 and 7 because the U exists as various species in solution depending on pH and thus its U adsorption mechanisms are different due to the U speciation. Based on experimental results and previous studies, the main U adsorption mechanisms of the WRK bentonite were understood in viewpoint of the chemical adsorption. At the acid conditions (＜pH 3), the U is apt to adsorb as forms of UO₂²⁺, mainly due to the ionic bond with Si-O or Al-O(OH) present on the WRK bentonite rather than the ion exchange with Ca²⁺ among layers of the WRK bentonite, showing the relatively low U adsorption efficiency. At the alkaline conditions (>pH 7), the U could be adsorbed in the form of anionic U-hydroxy complexes (UO₂(OH)₃^-, UO₂(OH)₄^2-, (UO₂)₃(OH)₇^-, etc.), mainly by bonding with oxygen (O^-) from Si-O or Al-O(OH) on the WRK bentonite or by co-precipitation in the form of hydroxide, showing the high U adsorption. At pH 7, the relatively low U adsorption efficiency (42%) was acquired in this study and it was due to the existence of the U-carbonates in solution, having relatively high solubility than other U species. The U adsorption efficiency of the WRK bentonite can be increased by maintaining a neutral or highly alkaline condition because of the formation of U-hydroxyl complexes rather than the uranyl ion (UO₂²⁺) in solution,and by restraining the formation of U-carbonate complexes in solution.

• Pediatric Gastroenterology, Hepatology & Nutrition
• /
• v.1 no.1
• /
• pp.56-78
• /
• 1998

Objective: Chronic hepatitis B infection (CHB) occurs in 6% to 10% of population in Korea. In ethinic communities where prevalence of chronic infection is high such as Korea, transmission of hepatitis B infection is either vertical (ie, by perinatal infection) or by close family contact (usually from mothers or siblings) during the first 5 years of life. The development of chronic hepatitis B infection is increasingly more common the earlier a person is exposed to the virus, particularly in fetal and neonatal life. And it progress to cirrhosis and hepatocellular carcinoma, especially in severe liver damage and perinatal infection. Histopathology of CHB is important when evaluating the final outcomes. A numerical scoring system which is a semiquantitatively assessed objective reproducible classification of chronic viral hepatitis, is a valuable tool for statistical analysis when predicting the outcome and evaluating antiviral and other therapies. In this study, a numerical scoring system (Ludwig system) was applied and compared with the conventional histological classification of De Groute. And the comparative analysis of cinical findings, family history, serology, and liver function test by histopathological findings in chronic hepatitis B of children was done. Methods: Ninety nine patients [mean age=9 years (range=17 months to 16 years)] with clinical, biochemical, serological and histological patterns of chronic HBV infection included in this study. Five of these children had hepatocelluar carcinoma. They were 83 male and 16 female children. They all underwent liver biopsies and histologic evaluation was performed by one pathologist. The biopsy specimens were classified, according to the standard criteria of De Groute as follows: normal, chronic lobular hepatitis (CLH), chronic persistent hepatitis (CPH), mild to severe chronic active hepatitis (CAH), or active cirrhosis, inactive cirrhosis, hepatocellular carcinoma (HCC). And the biopsy specimens were also assessed and scored semiquantitatively by the numerical scoring Ludwig system. Serum HBsAg, anti-HBs, HBeAg, anti-HBe, anti-HBc (IgG, IgM), and HDV were measured by radioimunoassays. Results: Male predominated in a proportion of 5.2:1 for all patients. Of 99 patients, 2 cases had normal, 2 cases had CLH, 22 cases had CPH, 40 cases had mild CAH, 19 cases had moderate CAH, 1 case had severe CAH, 7 cases had active cirrhosis, 1 case had inactive cirrhosis, and 5 cases had HCC. The mean age, sex distribution, symptoms, signs, and family history did not differ statistically among the different histologic groups. The numerical scoring system was correlated well with the conventional histological classification. The histological activity evaluated by both the conventional classification and the scoring system was more severe as the levels of serum aminotransferases were higher. In contrast, the levels of serum aminotransferases were not useful for predicting the degree of histologic activity because of its wide range overlapping. When the histological activity was more severe and especially the cirrhosis more progressing, the prothrombin time was more prolonged. The histological severity was inversely related with the duration of seroconversion of HBeAg. Conclusions: The histological activity could not be accurately predicted by clinical and biochemical findings, but by the proper histological classification of the numerical scoring system for the biopsy specimen. The numerical scoring system was correlated well with the conventional histological classification, and it seems to be a valuable tool for the statistical analysis when predicting the outcome and evaluating effects of antiviral and other therapies in chronic hepatitis B in children.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.227-252
• /
• 2018

The evolution of instant communication has mirrored the development of the Internet and messenger applications are among the most representative manifestations of instant communication technologies. In messenger applications, senders use emoticons to supplement the emotions conveyed in the text of their messages. The fact that communication via messenger applications is not face-to-face makes it difficult for senders to communicate their emotions to message recipients. Emoticons have long been used as symbols that indicate the moods of speakers. However, at present, emoticon-use is evolving into a means of conveying the psychological states of consumers who want to express individual characteristics and personality quirks while communicating their emotions to others. The fact that companies like KakaoTalk, Line, Apple, etc. have begun conducting emoticon business and sales of related content are expected to gradually increase testifies to the significance of this phenomenon. Nevertheless, despite the development of emoticons themselves and the growth of the emoticon market, no suitable emoticon recommendation system has yet been developed. Even KakaoTalk, a messenger application that commands more than 90% of domestic market share in South Korea, just grouped in to popularity, most recent, or brief category. This means consumers face the inconvenience of constantly scrolling around to locate the emoticons they want. The creation of an emoticon recommendation system would improve consumer convenience and satisfaction and increase the sales revenue of companies the sell emoticons. To recommend appropriate emoticons, it is necessary to quantify the emotions that the consumer sees and emotions. Such quantification will enable us to analyze the characteristics and emotions felt by consumers who used similar emoticons, which, in turn, will facilitate our emoticon recommendations for consumers. One way to quantify emoticons use is metadata-ization. Metadata-ization is a means of structuring or organizing unstructured and semi-structured data to extract meaning. By structuring unstructured emoticon data through metadata-ization, we can easily classify emoticons based on the emotions consumers want to express. To determine emoticons' precise emotions, we had to consider sub-detail expressions-not only the seven common emotional adjectives but also the metaphorical expressions that appear only in South Korean proved by previous studies related to emotion focusing on the emoticon's characteristics. We therefore collected the sub-detail expressions of emotion based on the "Shape", "Color" and "Adumbration". Moreover, to design a highly accurate recommendation system, we considered both emotion-technical indexes and emoticon-emotional indexes. We then identified 14 features of emoticon-technical indexes and selected 36 emotional adjectives. The 36 emotional adjectives consisted of contrasting adjectives, which we reduced to 18, and we measured the 18 emotional adjectives using 40 emoticon sets randomly selected from the top-ranked emoticons in the KakaoTalk shop. We surveyed 277 consumers in their mid-twenties who had experience purchasing emoticons; we recruited them online and asked them to evaluate five different emoticon sets. After data acquisition, we conducted a factor analysis of emoticon-emotional factors. We extracted four factors that we named "Comic", Softness", "Modernity" and "Transparency". We analyzed both the relationship between indexes and consumer attitude and the relationship between emoticon-technical indexes and emoticon-emotional factors. Through this process, we confirmed that the emoticon-technical indexes did not directly affect consumer attitudes but had a mediating effect on consumer attitudes through emoticon-emotional factors. The results of the analysis revealed the mechanism consumers use to evaluate emoticons; the results also showed that consumers' emoticon-technical indexes affected emoticon-emotional factors and that the emoticon-emotional factors affected consumer satisfaction. We therefore designed the emoticon recommendation system using only four emoticon-emotional factors; we created a recommendation method to calculate the Euclidean distance from each factors' emotion. In an attempt to increase the accuracy of the emoticon recommendation system, we compared the emotional patterns of selected emoticons with the recommended emoticons. The emotional patterns corresponded in principle. We verified the emoticon recommendation system by testing prediction accuracy; the predictions were 81.02% accurate in the first result, 76.64% accurate in the second, and 81.63% accurate in the third. This study developed a methodology that can be used in various fields academically and practically. We expect that the novel emoticon recommendation system we designed will increase emoticon sales for companies who conduct business in this domain and make consumer experiences more convenient. In addition, this study served as an important first step in the development of an intelligent emoticon recommendation system. The emotional factors proposed in this study could be collected in an emotional library that could serve as an emotion index for evaluation when new emoticons are released. Moreover, by combining the accumulated emotional library with company sales data, sales information, and consumer data, companies could develop hybrid recommendation systems that would bolster convenience for consumers and serve as intellectual assets that companies could strategically deploy.