• Journal of Korea Multimedia Society
• /
• v.18 no.7
• /
• pp.873-885
• /
• 2015

Today, It is difficult to search the various and numerous information efficiently. For this reason, Semantic Web emerged to provide searching services more easily through the structuring of a variety of unstructured format data and the definition of meaningful relationships between information. Especially, definition of relationship and meaning among resources is significant to share and infer related information. Ontology modeling plays just that role. Weapon parts development information is unstructured and dispersed all over. There are many difficulties in finding desired information, leading to getting improper outcomes. In this paper, we present an intuitive ontology model with weapon parts development information including the multi-dimensional information analysis and expansion of the relevant information. This study build up a ontology model through creating class and hierarchy about parts information and defining the properties of classes with Ontology Development 101[1] procedures using Protégé tools. The ontology model provides users with a platform on which search of needed information can be easy and efficient.

• Proceedings of the Korean Information Science Society Conference
• /
• 1998.10b
• /
• pp.508-510
• /
• 1998

기존의 절차 지향 프로그램을 이해하고 유지.보수하기 위해서는 많은 비용이 필요하다. 이러한 절차 지향 프로그램에서 객체/클래스를 추출한다면 프로그램을 이해하고 유지.보수하는데 많은 비용을 절감할 수 있을 뿐 아니라, 객체 지향 프로그램으로 변환하는데 많은 도움이 된다. 본 논문에서는 객체/클래스를 추출하기 위한 절차와 구조를 제시하였다. 객체/클래스 추출기는 Clustering Engine, Stochastic Refinement and Decision Engine, Domain Modelling와 Comparison and Intergration Engine의 4부분으로 구성된다. 이러한 과정을 거치면서 기존의 연구 방법과는 다르게 가중치 주는 기준, 다중 객체 후보, 통계적 방법으로의 정재와 결정, 요구사항의 의미적 관점에 기초한 방법을 사용하였다.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2003.05b
• /
• pp.476-479
• /
• 2003

전자문서 환경에서의 annotation은 그 특성상 원본문서의 내용이 변경될 경우 annotation의 대상인 앵커를 더 이상 참조할 수 없게 된다. 따라서 annotation 시스템에서는 반드시 원본문서 변경에 대한 앵커링 기능을 필요로 한다. 그러나 기존 연구에서는 앵커 텍스트의 변경을 고려하지 않거나 일반 텍스트 문서만을 대상으로 한다. 본 논문에서는 XML과 같은 구조문서 환경에서의 annotation 앵커링 기법을 제안한다. 제안된 기법에서는 XML 환경에서 앵커 텍스트 및 path정보에 대한 단계별 앵커링 과정을 수행한다. 또한 본 논문에서는 제안된 기법에 근거한 사용자 인터페이스를 제공한다. 그 결과 제안된 기법 및 시스템에서는 구조문서 환경에서 기존 연구 보다 심도 있는 앵커링을 보장하며 동시에 IETM, cyber-class, eLearing, semantic web 등의 다양한 분야에 효과적으로 적용 가능하다.

• Annual Conference of KIPS
• /
• 2010.11a
• /
• pp.640-643
• /
• 2010

영어권에서는 통계적 기반 의미 오류 검사기에 대한 연구가 활발하게 진행되었으나 한국어에서는 자주 나타나는 오류를 중심으로 오류 단어를 검사할 문맥이나 공기정보를 갖는 규칙 기반의 연구가 이루어져왔다. 본 논문에서는 표제어나 사전 기반 범주화 정보로 표현되고 있는 기존 문법 검사기의 성능을 개선하는 방안으로 한국어 명사 어휘 의미망인 KorLex를 활용하는 방법을 연구한다. 특히, 용언은 목적어나 주어에 사용되는 명사의 표제어가 의미 제약 조건으로 사용되는 예가 많다. 본 논문에서는 용언의 의미 제약 조건을 명사 표제어 단위에서 명사 의미 그룹(class) 단위로 확대하여 문법 검사기의 의미 오류 검사 규칙의 의미 오류 제약 조건을 일반화 하는 방안을 연구한다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.67-70
• /
• 2022

NLP 분야에서 적극 활용되기 시작한 Transformer 모델을 Vision 분야에서 적용하기 시작하면서 object detection과 segmentation 등 각종 분야에서 기존 CNN 기반 모델의 정체된 성능을 극복하며 향상되고 있다. 또한, label 데이터 없이 이미지들로만 자기지도학습을 한 ViT(Vision Transformer) 모델을 통해 이미지에 포함된 여러 중요한 객체의 영역을 검출하는 saliency map을 추출할 수 있게 되었으며, 이로 인해 ViT의 자기지도학습을 통한 object detection과 semantic segmentation 연구가 활발히 진행되고 있다. 본 논문에서는 ViT 모델 뒤에 classifier를 붙인 모델에 일반 학습한 모델과 자기지도학습의 pretrained weight을 사용해서 전이학습한 모델의 시각화를 통해 각 saliency map들을 비교 분석하였다. 이를 통해, 클래스 분류 학습 기반 전이학습이 transformer의 saliency map에 미치는 영향을 확인할 수 있었다.

• Journal of Digital Contents Society
• /
• v.9 no.2
• /
• pp.285-293
• /
• 2008

In Semantic Web, metadata and ontology for representing semantics and conceptual relationships of information resources are essential factors. RDF and RDF Schema are W3C standard models for describing metadata and ontology. Therefore, many studies to store and retrieve RDF and RDF Schema documents are required. In this paper, we focus on some results of analyzing available query patterns considering both RDF and RDF Schema and classify queries on RDF and RDF Schema into the three patterns. RDF and RDF Schema can be represented as graph models. So, we proposed some strategies to store and retrieve using the graph models of RDF and RDF Schema. We can retrieve entities that can be arrived from a certain class or property in RDF and RDF Schema without a loss of performance on account of multiple joins with tables.

• Journal of Science Education
• /
• v.37 no.3
• /
• pp.573-582
• /
• 2013

This study was conducted to inquire the semantic structure with science terminology used in middle school science class, and based on this, we wanted to look for the way to increase effectiveness of science teaching. In this study, we extracted twenty-six science terminologies used in "Force and Motion" unit in middle school science textbook, and administered word association test using the 26 science terminologies to 316 middle school students. As the result, we found that students had a divergent semantic structure to given science terminology, and there were cases to be interpreted as different meaning with teacher's intention. Also, we identified the terminologies which were not familiar to middle school students. It was found that female students were more familiar with science termilology than male students, and there were differences between schools.

• Proceedings of the Korea Contents Association Conference
• /
• 2006.05a
• /
• pp.231-234
• /
• 2006

Semantic web ontologies can be viewed as logic-based domain-oriented contents which allow distributed and heterogeneous information to be semantically integrated, automatically circulated, and enable implicit knowledge to be reasoned. This paper describes the 'Science and Technology Research Area' ontology which is being developed by the Korea Institute of Science and Technology Information (KISTI). This ontology was defined to assist actual researchers and project planners to grasp the researchers community from a variety of viewpoints. We describe classes and properties as ontology components and exemplify the representation of real instances in the ontology. In order to represent the identities of real world instances within the ontology, the above ontology employs both class-dependent URI assignment schemes and the identity resolution methods.

• Korean Journal of Logic
• /
• v.8 no.2
• /
• pp.109-141
• /
• 2005

In this paper, I raise an objection to "sensitivism" about "know", according to which knowledge ascription is sensitive to contexts of utterance or subjects. While Peter Unger once proposed insensitivism about "know" in terms of insensitivism with respect to absolute terms, David Lewis provided sensitivism about "know" in terms of sensitivism with respect to absolute terms, on the common ground that "know" belongs to a class of absolute terms. On the one hand, I object to Unger-style insensitivism about 'know,' for, I claim, we have reason to opt for sensitivism rather than insensitivism with respect to absolute terms in virtue of the maxim that I call "semantic razor." On the other hand, I also object to sensitivist approaches to "know," for, on reflection, there is such a deep difference between "know" and absolute terms (or, sensitive terms altogether) that "know" cannot be taken to sensitive to contexts as opposed to absolute terms (or, sensitive terms altogether). These claims jointly indicate that "know" should be thought of as an insensitive term even though sensitivism has enjoyed wide acceptance in many other cases.

• The Journal of Korean Association of Computer Education
• /
• v.9 no.6
• /
• pp.29-39
• /
• 2006

The Web is evolving quantitatively into an explosive development. However, users usually have heavy burden of searching information because of the absence of contextual meaning on the Web. Due to an enormous amount of information, users have to endure for finding strong cohesive keywords by themselves and read each of the documents with enduring effort. This paper proposes an efficient method of searching more relative documents than current KEM-based searching systems on the Web by using contextual meaning. We designed a domain ontology on computer hardware, and a searching system which was searching those e-Learning objects. Owing to the Ontology-applied search system, information such as educational materials and related multimedia can be easily provided to the users. Further, learners could be informed of relationship of knowledge, e.g., class hierarchy, properties and values, and so on. The request results are semantically related to users' needs, and thus the system provides a learner-centered searching.