• Journal of KIISE:Software and Applications
• /
• v.26 no.10
• /
• pp.1204-1218
• /
• 1999

문서영상 검색을 위한 디지털도서관의 대부분은 논문제목과/또는 논문요약으로부터 만들어진 색인에 근거한 제한적인 검색기능을 제공하고 있다. 본 논문에서는 영문 문서영상전체에 대한 검색을 위한 단어 영상 형태 특징기반의 단어검색시스템을 제안한다. 본 논문에서는 검색의 효율성과 정확도를 높이기 위해 1) 기존의 단어검색시스템에서 사용된 특징들을 조합하여 사용하며, 2) 특징의 개수 및 위치뿐만 아니라 특징들의 순서를 포함하여 매칭하는 방법을 사용하며, 3) 특징비교에 의해 검색결과를 얻은 후에 여과목적으로 문자인식을 부분적으로 적용하는 2단계의 검색방법을 사용한다. 제안된 시스템의 동작은 다음과 같다. 문서 영상이 주어지면, 문서 영상 구조가 분석되고 단어 영역들의 조합으로 분할된다. 단어 영상의 특징들이 추출되어 저장된다. 사용자의 텍스트 질의가 주어지면 이에 대응되는 단어 영상이 만들어지며 이로부터 영상특징이 추출된다. 이 참조 특징과 저장된 특징들과 비교하여 유사한 단어를 검색하게 된다. 제안된 시스템은 IBM-PC를 이용한 웹 환경에서 구축되었으며, 영문 문서영상을 이용하여 실험이 수행되었다. 실험결과는 본 논문에서 제안하는 방법들의 유효성을 보여주고 있다. Abstract Most existing digital libraries for document image retrieval provide a limited retrieval service due to their indexing from document titles and/or the content of document abstracts. This paper proposes a word spotting system for full English document image retrieval based on word image shape features. In order to improve not only the efficiency but also the precision of a retrieval system, we develop the system by 1) using a combination of the holistic features which have been used in the existing word spotting systems, 2) performing image matching by comparing the order of features in a word in addition to the number of features and their positions, and 3) adopting 2 stage retrieval strategies by obtaining retrieval results by image feature matching and applying OCR(Optical Charater Recognition) partly to the results for filtering purpose. The proposed system operates as follows: given a document image, its structure is analyzed and is segmented into a set of word regions. Then, word shape features are extracted and stored. Given a user's query with text, features are extracted after its corresponding word image is generated. This reference model is compared with the stored features to find out similar words. The proposed system is implemented with IBM-PC in a web environment and its experiments are performed with English document images. Experimental results show the effectiveness of the proposed methods.

• The Journal of the Convergence on Culture Technology
• /
• v.7 no.4
• /
• pp.325-330
• /
• 2021

This study set out to investigate the forms, attributes, and escape methods of post-subjects projected on the investigator's works in connection with rhizomatic thinking proposed as a way of social transformation by Deleuze and Guattari and examine their social connotations. Post-subjects projected on the investigator's works are not completed wholes of some sort, but like materials whose constant premise is change and creation. In the investigator's works, post-subjects have conscious and unconscious desire. It is the desire of creation with positive attributes including Deleuze's and Guattari's pursuit of changes in a contradicting society. When desire is deployed in post-subjects, they will carry out an escape. This way of escape is rhizomatic proposed by Deleuze and Guattari. It deconstructs contradicting things and repeats connection, contact, and severance with the outside world, building a new order. Rhizomatic post-subjects appearing in the investigator's works depict the escape process and method in abstract ways through the variable installation of objets combined with a color field of repeating brushes. In this work, the goal of post-subjects is to make a safe landing in a space where beings are recognized for their values and free and creative lives. These post-subjects are nomads creating a new landscape continuously, wandering around vast plains, and also artists and literary figures resisting a contradicting society. That is, they are connected to the concept of a war machine proposed by Deleuze and Guattari as a concept of social transformation and to the concept of Nietzsche's Agon to devise and create new values and politics based on street passion. They seek after a space where they can co-exist with otherness recognized rather than the complete deconstruction of the old order.

• Informatization Policy
• /
• v.31 no.2
• /
• pp.3-38
• /
• 2024

Hallucination is a significant barrier to the utilization of large-scale language models or multimodal models. In this study, we collected 654 computer science papers with "hallucination" in the abstract from arXiv from December 2022 to January 2024 following the advent of Chat GPT and conducted frequency analysis, knowledge network analysis, and literature review to explore the latest trends in hallucination research. The results showed that research in the fields of "Computation and Language," "Artificial Intelligence," "Computer Vision and Pattern Recognition," and "Machine Learning" were active. We then analyzed the research trends in the four major fields by focusing on the main authors and dividing them into data, hallucination detection, and hallucination mitigation. The main research trends included hallucination mitigation through supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF), inference enhancement via "chain of thought" (CoT), and growing interest in hallucination mitigation within the domain of multimodal AI. This study provides insights into the latest developments in hallucination research through a technology-oriented literature review. This study is expected to help subsequent research in both engineering and humanities and social sciences fields by understanding the latest trends in hallucination research.

• Journal of the Korean BIBLIA Society for library and Information Science
• /
• v.7 no.1
• /
• pp.5-22
• /
• 1994

This paper attempts to define some general characteristics of the Abstracting Journals published in Korea as evidenced in those published during last ten years. This purpose is achieved by comparing the results of the two studies conducted by the author in 1984 and in 1994. Both studies were conducted to present the state of the art in the abstracting services in Korea. The major conclusions made in this paper are summarised as follows： (1) Researchers and professionals working in a small number of subject fields are benefited by the abstracting journals, which provide current-awareness services of recent achievements in research and development in Korea. Those in most of the fields have no abstracting journals of their own, and naturally they have no substantial abstract-ing services. Even many researchers and professionals in the fields that have some abstracting journals are not informed of research results in their fields because the abstracting journals are scattered in many narrow subjects and in many cases, the abstracting journals only cover publications in some specific forms and kinds. (2) Abstracting journals that cover more than two subject fields, which are supposed to be of more or less help to the researchers and professionals in the subject fields that have no abstracting journals published in their fields, have rapidly increased in number in the past ten years. Most of suh abstracting journals carry thesis and dissertation abstracts, and the rest, those of research papers published in specific places, in specific forms, by specific institutions, and of reports of research projects sponsored by specific foundations. These abstracting journals are not of the kind that comprehensively provide researchers in related fields with current awareness of publications of research results in Korea. (3) Most of the abstracting Journals existing in Korea are Published by institutions of higher education and research institutes, and the rest, by commercial publishers, industrial firms, libraries, information centres, government agencies, research foundations, learned societies, etc. Those which publish many titles are small in number and those publish one or two titles are large in number. The former is largely made up of institutions of higher education and research institutes. (4) The abstracting journals published in Korea are classified by type into those of dissertations, research papers, journal articles, patent specifications in that descending order. The fact that Master; and doctoral dissertation abstracts ate dominating in Korea is due to the irrational practice of publishing those abstracts at many different institutions. (5) Most of the abstracting journals existing in Korea are published by national or government-supported research institutes in order to publicise their own research outputs. Their coverage of literature is normally narrow, and naturally their value to users is limited. (6) Korean is the desirable language for the abstracting journals intended to be distributed within Korea. About half of the abstracting jornals published in Korea is printed in Korean and the other half, in foreign languages, and in Korean and in foreign languages together. All the abstracting journals in foreign languages are printed in English except one, which is printed in Japanese. (7) Some twenty per cent of the abstracting journals in Korea is published monthly, bimonthly, and quarterly. The others are published annually, biannually and irregularly. The latter may not function properly as a current-awareness tool due to long intervals between their issues. It is particularly undesirable that about half of the abstracting journals in Korea is published irregularly. Most of the abstracting journals published in Korea are distributed freely to individuals and institutions selected by the publishers. (8) The abstracting journals published by the use of computers increased drastically in the past ten years. The abstracting journals produced by the conventional type-setting method will possibly disappear in Korea in another ten years to come. Automation of the production of abstracting journals does not simply mean technical, economic improvement in publishing processes but availability of machine-readable databases that can be used for many other pur-poses, including generation of other bibliographical publications and provision of machine literature searching capabilities. Necessary steps should be taken for this important development immediately.

• Journal of Internet Computing and Services
• /
• v.18 no.1
• /
• pp.77-88
• /
• 2017

Recently, with the advent of knowledge based society where information and knowledge make values, patents which are the representative form of intellectual property have become important, and the number of the patents follows growing trends. Thus, it needs to classify the patents depending on the technological topic of the invention appropriately in order to use a vast amount of the patent information effectively. IPC (International Patent Classification) is widely used for this situation. Researches about IPC automatic classification have been studied using data mining and machine learning algorithms to improve current IPC classification task which categorizes patent documents by hand. However, most of the previous researches have focused on applying various existing machine learning methods to the patent documents rather than considering on the characteristics of the data or the structure of patent documents. In this paper, therefore, we propose to use two structural fields, technical field and background, considered as having impacts on the patent classification, where the two field are selected by applying of the characteristics of patent documents and the role of the structural fields. We also construct multi-label classification model to reflect what a patent document could have multiple IPCs. Furthermore, we propose a method to classify patent documents at the IPC subclass level comprised of 630 categories so that we investigate the possibility of applying the IPC multi-label classification model into the real field. The effect of structural fields of patent documents are examined using 564,793 registered patents in Korea, and 87.2% precision is obtained in the case of using title, abstract, claims, technical field and background. From this sequence, we verify that the technical field and background have an important role in improving the precision of IPC multi-label classification in IPC subclass level.

• The KIPS Transactions:PartB
• /
• v.10B no.3
• /
• pp.281-286
• /
• 2003

A large number of features are collected for problem solving in real life, but to utilize ail the features collected would be difficult. It is not so easy to collect of correct data about all features. In case it takes advantage of all collected data to learn, complicated learning model is created and good performance result can't get. Also exist interrelationships or hierarchical relations among the features. We can reduce feature's number analyzing relation among the features using heuristic knowledge or statistical method. Heuristic technique refers to learning through repetitive trial and errors and experience. Experts can approach to relevant problem domain through opinion collection process by experience. These properties can be utilized to reduce the number of feature used in learning. Experts generate a new feature (highly abstract) using raw data. This paper describes machine learning model that reduce the number of features used in learning using heuristic function and use abstracted feature by neural network's input value. We have applied this model to the win/lose prediction in pro-baseball games. The result shows the model mixing two techniques not only reduces the complexity of the neural network model but also significantly improves the classification accuracy than when neural network and heuristic model are used separately.

• Journal of the Korean Society of Costume
• /
• v.50 no.7
• /
• pp.195-218
• /
• 2000

This study investigates the characteristics of European knitted works from a historical perspective. Specifically, this study deals with the following research topics: 1) the origin and development of knitting. 2) the characteristics of knitting industry according to the change of times, 3) the comparison of local knitting patterns and cultures. 4) 7he symbolic meaning of the designs in the knitted works and theire functions. This research is barred on the survey of the relevant literature and photographs. The results of the study are summarized as follows. 1) The introduction of knitted works was closely connected with the climatic and socio-economic conditions of the places of the origin. Knitted work developed mostly in Northern Europe, a cold area, and the barren, mountainous coastal areas where people frequently used woolen materials for clothes. 2) In ancient times, abstract and geometric patterns have developed in Europe under the influence of Arabian knitted work. Middle Ages saw the flourishing of Arabian knitted works representing the authority of the church. In early modern times, the knitted work assumed the wealth of the royal families and the nobles. But afterward it was gradually Popularized among the middle classes. Knitting was then regarded as one of the women's major cultural activities. However, recently in the interwar periods. the knitting industry did not flourish and the knitted works came to serve merely as comfort goods by political urge. Knitted works were introduced in Korea around 1870 (the 7th or 8th year of king Kojong era) by Catholic missionaries and they started to be made by machine in 1917. 3) As for the propagation of the knitted work into Europe, there are three routes estimated. The traditional knitting patterns of local areas and their characteristics are summed up as follows : (1) England Guernseys are thick dark blue wool, whereas Jerseys are thinner and of various colors. The knitted shawls of Shetland are world-famous for their fine, lace-like texture that they can be through a wedding-ring. The knitted work of Fair Isle shows several distinctive features, such as the use of no more than two colors, patterns with diagonal lines. symmetry within the patterns, the prominent OXO patterns, and horizontal bands of patterning. The representative knitted work of Aran is Aran sweater made for fishermen to developed from guernseys of Scotland. (2) Scandinavian countries are distinguished from other countries by their conservative but creative cultural tradition. Their knitting patterns are characterized by small geometric figures such as dots, triangles, squares, rhombuses, and crosses used often with stars and roses. Scandinavian knitting is also salient for its vertical stripes and simple motifs repeating at short intervals. (3) Baltic area : The Latvian and Lithuania stockings have very ornate patterns. Many of the Estonian knit stockings and mittens share designs. Komi was well-known for its symmetric diamond pattern. Komi patterns include colored stripes, borders of pattern and all-over designs of complex diagonals. (4) Balkan area : In Yugoslavia, the patterns of roses, leaves and flowers were used for stockings, gloves and leggings. Greek knitting resembled southern Russian knitting, which utilized light colored patterns with dark colors for a background. Turkish patterns are symmetric vertically or horizontally. 4) The traditional knitting patterns net only carried symbolic meanings but also served as means of communication. First of all, patterns had incantatory meanings. Patterns also represented Power or authenticity Patterns were symbolic of one's social standing, too. The colors, motifs and their arrangements were very important features symbolizing one's social position or family line. People often communicated by certain pieces of knitted work or patterns.