• Proceedings of the Korean Information Science Society Conference
• /
• 2005.11b
• /
• pp.688-690
• /
• 2005

본 논문은 웹에 산재되어 있는 OWL/RDF(S) 도메인 온톨로지에 대한 검색 시스템을 설계하여, 온톨로지의 재사용성을 극대화 하는데 그 목적이 있다. 컴퓨터와 인간이 정보를 공유하고, 의미적인 상호작용을 위한 시맨틱 웹에 대한 연구가 활발이 진행되고 있다. 시맨틱 웹을 실현하기 위해 개념들의 정의와 개념들간의 관계를 형성하는 온톨로지의 구축이 필수요소가 됨에 따라 온톨로지를 구축하기 위해 OWL, RDF(S) 그리고 DAML+OIL 등의 많은 온롤로지 언어가 개발되었고, 이들 언어를 기반으로 하는 Protege, OILEd와 KAON 등의 사용자들에게 온톨로지 구축의 편리성을 제공하는 온톨로지 구축 도구들도 50여가지 이상 개발되었다. 이러한 이유로 많은 온톨로지들이 개발되고 있다. 그렇지만 온톨로지의 가장 큰 특징은 동일 도메인의 온톨로지의 재사용인데, 산재되어 있는 온톨로지들을 검색하기 어렵고, 이들을 한데 모아놓은 저장소 또한 갖추어지지 않아 동일한 도메인 온톨로지가 존재할지라도 새롭게 온톨로지를 구축해야한다. 이에 본 논문에서는 웹상에 존재하는 온톨로지들의 검색을 용이하게 하여 지식 정보의 재사용을 최대화하기 위하여 본 연구를 진행하고 시스템을 설계하였다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.10d
• /
• pp.292-294
• /
• 2002

대부분의 생물학에서는 많은 지식을 도출 할 수 있는 공리(axiom)에 의한 응용 프로그램보다는 기존의 지식을 적용하고 있으며, 바이오인포매틱스 데이터베이스에 저장된 복잡한 생물학 자료는 추가, 변경이 자주 발생한다. 이러한 바이오인포매틱스의 데이터베이스와 응용 프로그램에서 지식을 표현하는 방법으로써 온톨로지의 사용이 제시되고 있다. 온톨로지는 사람과 컴퓨터간의 공유되는 지식을 개념화하고, 이를 명세화 하는 것으로 정의된다. 즉, 온톨로지는 도메인 내의 지식을 개념화한 구체적인 형식이며, 개념화와 개념화간의 관계를 표현할 수 있다. 또한, W3C에서 제안한 시맨틱 웹은 온톨로지는 중요 기술로 사용하고 있으며, 온톨로지를 통한 추론과 컴퓨터가 이해 가능한 형식을 제공하여 상호운영성등을 향상시킨다. 본 논문에서는 기존의 바이오 온톨로지들에 대해서 알아보고, 바이오 온톨로지 시스템의 설계와 시스템의 각 구성 요소에 대해서 제시한다. 마지막으로 이러한 시스템을 구축할 때에 고려되어야 하는 이슈들에 대해서 설명한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2014.04a
• /
• pp.898-900
• /
• 2014

시맨틱 검색은 논리적으로 표현된 지식 베이스를 사용하여 현재의 키워드 기반 검색보다 더 정확한 결과를 제공할 수 있다. 그러나 일반 사용자는 지식 기반의 복잡하고 정형화된 질의어와 스키마를 잘 알지 못한다. 그래서 검색 시스템은 사용자 키워드의 의미를 해석할 수 있어야 한다. 본 논문에서는 멀티미디어 콘텐츠의 시맨틱 검색을 위한 사용자 질의 의미 해석 시스템을 설명한다. 제안한 시스템은 도메인 온톨로지 기반으로 구축된 지식 베이스의 정형화된 구조에 의미 해석 과정이 통합된 온롤로지 지식 베이스 기반 검색 시스템이다.

• Database Research
• /
• v.34 no.3
• /
• pp.137-147
• /
• 2018

A key to a well-performing human activity recognition (HAR) system through machine learning technique is the availability of a substantial amount of labeled data. Collecting sufficient labeled data is an expensive and time-consuming task. To build a HAR system in a new environment (i.e., the target domain) with very limited labeled data, it is unfavorable to naively exploit the data or trained classifier model from the existing environment (i.e., the source domain) as it is due to the domain difference. While traditional machine learning approaches are unable to address such distribution mismatch, transfer learning approach leverages the utilization of knowledge from existing well-established source domains that help to build an accurate classifier in the target domain. In this work, we propose a transfer learning approach to create an accurate HAR classifier with very limited data through the multitask neural network. The classifier loss function minimization for source and target domain are treated as two different tasks. The knowledge transfer is performed by simultaneously minimizing the loss function of both tasks using a single neural network model. Furthermore, we utilize the unlabeled data in an unsupervised manner to help the model training. The experiment result shows that the proposed work consistently outperforms existing approaches.

• Journal of the Korean Society for information Management
• /
• v.19 no.2
• /
• pp.109-130
• /
• 2002

This paper is mainly concerned with the matter of how organizational knowledge is represented in terms of usability by letting the members of the organization evaluate knowledge items in the problem solving situations. Organizational knowledge can be represented in the form of a visualized dynamic knowledge map since the organizational members experience constant changes in goals, kinds of problems, and the importance of alternatives to solve the given problems. The dynamic map suggested in this study is goal oriented in the sense that it represents the relationship between knowledge items across domains as well as within a domain.

• Journal of the Korean BIBLIA Society for library and Information Science
• /
• v.32 no.1
• /
• pp.151-170
• /
• 2021

The domain name should be registered and used with the minimum length in consideration of the user's web accessibility and convenience. In addition, it is necessary to increase the status and authority of the library by allowing the type, region, and characteristics of the library to be known only by the domain name, and to consider the convenience of users. In this study, the situation of the state designated second-stage public domain and the second-stage public domain of educational institutions of similar character and size to the library are analyzed, and problems with the domain of libraries that do not have the second-stage domain are identified. Finally, second-stage public domains of national library (nl), public library (pl), and small library (sl) are proposed for each type of library, and the university library (lib. university domain) and school library (lib. School domain) proposed to use as the second-stage domains of the educational institution. The purpose is to allow users to intuitively know that these domains are libraries, to identify the type of library, and to know the characteristics and regions of the library. To this end, a joint effort by the academic community, the library community, and the Library Information Policy Committee is needed.

• Journal of Advanced Navigation Technology
• /
• v.2 no.1
• /
• pp.61-67
• /
• 1998

In this paper, we have described the Ramp Activity Coordination Expert System (RACES) which can solve aircraft parking problems. RACES includes a knowledge-based scheduling problem which assigns every daily arriving and departing flight to the gates and remote spots with the domain specific knowledge and heuristics acquired from human experts. RACES processes complex scheduling problem such as dynamic inter-relations among the characteristics of remote spots/gates and aircraft with various other constraints, for example, custome and ground handling factors at an airport. By user-driven modeling for end users and knowledge-driven near optimal scheduling acquired from human experts, RACES can produce parking schedules of aircraft in about 20 seconds for about 400 daily flights, whereas it normally takes about 4 to 5 hours by human experts. Scheduling results in the form of Gantt charts produced by the RACES are also accepted by the domain experts. RACES is also designed to deal with the partial adjustment of the schedule when unexpected events occur. After daily scheduling is completed, the messages for aircraft changes and delay messages are reflected and updated into the schedule according to the knowledge of the domain experts. By analyzing the knowledge model of the domain expert, the reactive scheduling steps are effectively represented as rules and the scenarios of the Graphic User Interfaces (GUI) are designed. Since the modification of the aircraft dispositions such as aircraft changes and cancellations of flights are reflected to the current schedule, the modification should be notified to RACES from the mainframe for the reactive scheduling. The adjustments of the schedule are made semi-automatically by RACES since there are many irregularities in dealing with the partial rescheduling.

• Journal of Intelligence and Information Systems
• /
• v.28 no.4
• /
• pp.251-273
• /
• 2022

Recently, research on applying text analysis to deep learning has steadily continued. In particular, researches have been actively conducted to understand the meaning of words and perform tasks such as summarization and sentiment classification through a pre-trained language model that learns large datasets. However, existing pre-trained language models show limitations in that they do not understand specific domains well. Therefore, in recent years, the flow of research has shifted toward creating a language model specialized for a particular domain. Domain-specific pre-trained language models allow the model to understand the knowledge of a particular domain better and reveal performance improvements on various tasks in the field. However, domain-specific further pre-training is expensive to acquire corpus data of the target domain. Furthermore, many cases have reported that performance improvement after further pre-training is insignificant in some domains. As such, it is difficult to decide to develop a domain-specific pre-trained language model, while it is not clear whether the performance will be improved dramatically. In this paper, we present a way to proactively check the expected performance improvement by further pre-training in a domain before actually performing further pre-training. Specifically, after selecting three domains, we measured the increase in classification accuracy through further pre-training in each domain. We also developed and presented new indicators to estimate the specificity of the domain based on the normalized frequency of the keywords used in each domain. Finally, we conducted classification using a pre-trained language model and a domain-specific pre-trained language model of three domains. As a result, we confirmed that the higher the domain specificity index, the higher the performance improvement through further pre-training.

• Journal of the KSME
• /
• v.57 no.8
• /
• pp.49-53
• /
• 2017

이 글에서는 스마트 팩토리 구현을 위해 고려되어야 하는 다양한 속성들 중에서 '데이터'에 대해서 논하고자 한다. 특히, 제조업의 관점에서 데이터의 가치를 평가하는 방법과 도메인 지식의 중요성에 대해 설명하였고, 중소기업에서 데이터를 활용하기 위한 방안을 함께 고민해보았다.

• Journal of the Korea Society of Computer and Information
• /
• v.11 no.3
• /
• pp.87-96
• /
• 2006

Most conventional database systems support specific queries that are concerned only with data that match a query qualification precisely. A cooperative query answering supports query analysis, query relaxation and provides approximate answers as well as exact answers. The key problem in the cooperative answering is how to provide an approximate functionality for alphanumeric as well as categorical queries. In this paper, we propose a metricized knowledge abstraction hierarchy that supports multi-level data abstraction hierarchy and distance metric among data values. In order to facilitate the query relaxation, a knowledge representation framework has been adopted, which accommodates semantic relationships or distance metrics to represent similarities among data values. The numeric domains also compatibly incorporated in the knowledge abstraction hierarchy by calculating the distance between target record and neighbor records.