• The Korean Journal of Applied Statistics
• /
• v.34 no.3
• /
• pp.389-399
• /
• 2021

In recent years, word embedding has been a popular field of natural language processing research and a skip-gram has become one successful word embedding method. It assigns a word embedding vector to each word using contexts, which provides an effective way to analyze text data. However, due to the limitation of vector space model, primary word embedding methods assume that every word only have a single meaning. As one faces multi-sense words, that is, words with more than one meaning, in reality, Neelakantan (2014) proposed a multi-sense skip-gram (MSSG) to find embedding vectors corresponding to the each senses of a multi-sense word using a clustering method. In this paper, we propose a modified method of the MSSG to improve statistical accuracy. Moreover, we propose a data-adaptive choice of the number of clusters, that is, the number of meanings for a multi-sense word. Some numerical evidence is given by conducting real data-based simulations.

• Smart Media Journal
• /
• v.11 no.8
• /
• pp.9-20
• /
• 2022

With the rise of the metaverse industry worldwide, relevant national strategies and nurturing systems have been prepared in Korea. As the complexity of policies increases, the importance of establishing data-based policymkaing is growing, and studies diagnosing national R&D trends in the metaverse field are still lacking. Therefore, this paper collected NTIS national R&D information for 9,651 R&D projects promoted from 2002 to 2020. And this study looked at the current status and identified major topics based on the topic modeling, and considered time-series changes in the topics. Eleven major topics of R&D tasks in the metaverse field were derived, hot topics were service/content/platform development and medical/surgical fields of application fields, and cold topics were urban/environment/spatial information fields. Strategic R&D Management, metaverse-related laws, and institutional studies were proposed as policy directions.

• Journal of Industrial Convergence
• /
• v.22 no.3
• /
• pp.13-25
• /
• 2024

This study explores research topics related to dance performances published in Korea based on big data and examines research trends that change according to the trend of the times. The results derived from topic modeling analysis are as follows. (1) Six major topics were derived: a study on marketing strategies and development plans for dance performances, (2) a study on the re-watching factors of dance performance space and performance satisfaction, (3) a study on the popularity and contribution of dance performances in the stage environment, (4) a study on the current status of dance performances and the convergence of dance group operations, (5) a study on the definition of dance performances using various social media, and (6) a study on the direction and development of technology-applied dance performance contents. Accordingly, research trends and topics related to dance, including dance performances, social changes, key keywords of researchers' change interests were extracted, and keywords were compared and analyzed to present academic changes and countermeasures. Accordingly, the need for research to apply new technologies was emphasized as it diversified and fused.

• Convergence Security Journal
• /
• v.23 no.4
• /
• pp.61-70
• /
• 2023

Clustering is an unsupervised learning method that involves grouping data based on features such as distance metrics, using data without known labels or ground truth values. This method has the advantage of being applicable to various types of data, including images, text, and audio, without the need for labeling. Traditional clustering techniques involve applying dimensionality reduction methods or extracting specific features to perform clustering. However, with the advancement of deep learning models, research on deep clustering techniques using techniques such as autoencoders and generative adversarial networks, which represent input data as latent vectors, has emerged. In this study, we propose a deep clustering technique based on deep learning. In this approach, we use an autoencoder to transform the input data into latent vectors, and then construct a vector space according to the cluster structure and perform k-means clustering. We conducted experiments using the MNIST and Fashion-MNIST datasets in the PyTorch machine learning library as the experimental environment. The model used is a convolutional neural network-based autoencoder model. The experimental results show an accuracy of 89.42% for MNIST and 56.64% for Fashion-MNIST when k is set to 10.

• Journal of the Korean Institute of Landscape Architecture
• /
• v.49 no.2
• /
• pp.51-60
• /
• 2021

Modern society is becoming more informed and intelligent with the development of digital technology, in which humans, objects, and networks relate with each other. In accordance with the changing times, a garden system has emerged that makes it easy to supply the ideal temperature, humidity, sunlight, and moisture conditions to grow plants. Therefore, this study attempted to grasp the concept, perception, and trends of smart gardens, a recent concept. To achieve the purpose of this study, previous studies and text mining were used, and the results are as follows. First, the core characteristics of smart gardens are new gardens in which IoT technology and gardening techniques are fused in indoor and outdoor spaces due to technological developments and changes in people's lifestyles. As technology advances and the importance of the environment increases, smart gardens are becoming a reality due to the need for living spaces where humans and nature can co-exist. With the advent of smart gardens, it will be possible to contribute to gardens' vitalization to deal with changes in garden-related industries and people's lifestyles. Second, in current research related to smart gardens and users' experiences, the technical aspects of smart gardens are the most interesting. People value smart garden functions and technical aspects that enable a safe, comfortable, and convenient life, and subjective uses are emerging depending on individual tastes and the comfort with digital devices. Third, looking at the usage behavior of smart gardens, they are mainly used in indoor spaces, with edible plants are being grown. Due to the growing importance of the environment and concerns about climate change and a possible food crisis, the tendency is to prefer the cultivation of plants related to food, but the expansion of garden functions can satisfying users' needs with various technologies that allow for the growing of flowers. In addition, as users feel the shapes of smart gardens are new and sophisticated, it can be seen that design is an essential factor that helps to satisfy users. Currently, smart gardens are developing in terms of technology. However, the main components of the smart garden are the combination of humans, nature, and technology rather than focusing on growing plants conveniently by simply connecting potted plants and smart devices. It strengthens connectivity with various city services and smart homes. Smart gardens interact with the landscape of the architect's ideas rather than reproducing nature through science and technology. Therefore, it is necessary to have a design that considers the functions of the garden and the needs of users. In addition, by providing citizens indoor and urban parks and public facilities, it is possible to share the functions of communication and gardening among generations targeting those who do not enjoy 'smart' services due to age and bridge the digital device and information gap. Smart gardens have potential as a new landscaping space.

• The KIPS Transactions:PartB
• /
• v.11B no.6
• /
• pp.749-758
• /
• 2004

In this paper, we propose a new method utilizing only raw corpus without additional human effort for disambiguation of target word selection in English-Korean machine translation. We use two data-driven techniques; one is the Latent Semantic Analysis(LSA) and the other the Probabilistic Latent Semantic Analysis(PLSA). These two techniques can represent complex semantic structures in given contexts like text passages. We construct linguistic semantic knowledge by using the two techniques and use the knowledge for target word selection in English-Korean machine translation. For target word selection, we utilize a grammatical relationship stored in a dictionary. We use k- nearest neighbor learning algorithm for the resolution of data sparseness Problem in target word selection and estimate the distance between instances based on these models. In experiments, we use TREC data of AP news for construction of latent semantic space and Wail Street Journal corpus for evaluation of target word selection. Through the Latent Semantic Analysis methods, the accuracy of target word selection has improved over 10% and PLSA has showed better accuracy than LSA method. finally we have showed the relatedness between the accuracy and two important factors ; one is dimensionality of latent space and k value of k-NT learning by using correlation calculation.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.9 no.5
• /
• pp.183-189
• /
• 2009

Recently, PCS, PDA and mobile devices, such as the proliferation of spread, GPS (Global Positioning System) the use of, the rapid development of wireless network and a regular user even images, audio, video, multimedia data, such as increased use is for. In particular, video data among multimedia data, unlike the moving object, text or image data that contains information about the movements and changes in the space of time, depending on the kinds of changes that have sigongganjeok attributes. Spatial location of objects on the flow of time, changing according to the moving object (Moving Object) of the continuous movement trajectory of the meeting is called, from the user from the database that contains a given query trajectory and data trajectory similar to the finding of similar trajectory Search (Similar Sub-trajectory Retrieval) is called. To search for the trajectory, and these variations, and given the similar trajectory of the user query (Tolerance) in the search for a similar trajectory to approximate data matching (Approximate Matching) should be available. In addition, a large multimedia data from the database that you only want to be able to find a faster time-effective ways to search different from the existing research is required. To this end, in this paper effectively divided into a grid to search for the trajectory to the trajectory of moving objects, similar to the effective support of the search trajectory offers a new grid-based search techniques.

• The Journal of the Korea Contents Association
• /
• v.16 no.9
• /
• pp.779-790
• /
• 2016

This study is to exhibit the iconographic narrative and visual culture of ancient Chinese art. The focus of the study is the composite integration of literature and graphic forms, in particular the heterochronous expression of different scenarios of scenes occurring in different time periods in pictures of ancient art. The unity of their origins with picture narration and comic art creation is the fusion of our modern times. The ancient Chinese understanding of visual art includes the traditional style of images and their symbolic meanings. Among artistic narrative expression, imagery contemplation and visual presentation have significance. Artistic thinking is inseparable from visual articulation. It is a rational thought process through creative language interpretation in visual media of imagery narratives. The characteristics of ancient imagery thinking and the way of presenting sequential incidents in the form pictures is a creative space of time. This is the spatial thinking of modern comic art, which is demonstrated through acceptance in artistic styles. Image narration needs new forms and media styles, including integrating with cultural values as aesthetic communication is necessary.

• Archives of design research
• /
• v.19 no.6 s.68
• /
• pp.27-32
• /
• 2006

TYME is an interactive typography program based on the ephemeral nature of time. It is a text based poetic tool, built with Processing. This project is presented as a performance that utilizes a computer with a display screen, a projector, and speakers. As the user types, white characters appear, flow on the score on the screen and trigger jazz sounds, then characters disappear into black space in several seconds on the screen like smoke. Typography from this invented instrument is evocative and wistful, and allows the user to associate with ephemeral time. While typing the characters as though playing an instrument at intervals of time, the user can freeze the motion and print out or save otherwise fleeting moment with a button. I intend to contain both characteristics: the amorphous shape of smoke and the elusive attribute of smoke for the expression of intangible and ephemeral time. Every alphabetic shape is derived from the video dips that I shot of smoke. The resulting alphabetic images are then programmed using the Processing scripting language and which can then be typed on the screen with a keyboard. TYME could be a model as a project that reflects the unfixed quality of digital typography, and as a design approach for interactive expressive typography by scripting code. This project also represents the characteristics of typographic play, which can be realized in an computational environment like this model.

• Journal of Korean Society for Geospatial Information Science
• /
• v.15 no.4
• /
• pp.59-69
• /
• 2007

The scale of a natural disaster grows bigger and bigger every year. The government spends much of its budget on recovering the resulting damage on a national scale. It is important to shift the paradigm from taking measures after a disaster to that of taking preventive actions before a disaster in order to bring a fundamental resolution to such problems. In taking preventive actions and policies, it is necessary to integrate various kinds of advanced technologies including IT, high-tech information gathering technology and operational technology, and to predict and evaluate natural disasters on a comprehensive level. Although Korea is a country with a strong IT sector, most information gathering is still performed in paper. In particular, information about the areas of previous landslide occurrences and slopes remains on paper, which makes it difficult to share the information and to discern the contents, and also raises the possibility of missing documents. Thus this study set out to develop an information gathering and management system for cutting slopes using PDA from the perspective of information gathering, system compatibility, and information management. As a result, field information may be gathered in a variety of forms (location, photos, and texts) real-time. A rough judgment was also made of the stability of rock slopes using the SMR method on the field.