• Proceedings of the Korea Information Processing Society Conference
• /
• 2014.04a
• /
• pp.798-801
• /
• 2014

문서에서 공간 개체와 사건을 찾아내고, 이들 간의 위상적 관계나 의미적 관계를 찾아내는 것을 공간정보 추출이라고 한다. 본 논문에서는 언어분석 결과와 세종사전을 활용해 자연언어 문서에서 동작(motion) 사건 관계 중심의 공간 정보를 추출하는 규칙 기반 시스템을 제안하였다. 수동으로 구축한 20문장의 평가 집합에 대해 사건 관계 추출은 27.45%의 F-measure 성능을 보였다. 공간보다 비교적 많은 연구가 진행된 시간 관계 추출에 대한 최신 연구의 성능이 30~35% 수준[1]인 것을 고려하여 볼 때, 본 연구는 공간 사건 관계 추출의 기초 연구로 의미가 있다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.04a
• /
• pp.819-821
• /
• 2015

문맥 철자 오류란 단어만 봤을 때에는 오류가 아니지만 문맥상으로는 오류인 문제를 말한다. 이 문제를 해결하기 위해서는 문맥 정보를 보아야 하는데 기존의 방법들은 언어학의 전문가가 설계한 규칙을 사용하거나, 통계적인 분석 방법을 사용하였다. 하지만 이 방법들은 많은 시간과 노력을 필요로 하지만 높은 성능을 얻지 못한다. 본 논문에서는 최근 자연언어처리에서 연구되고 있는 딥러닝을 사용하여 문맥 철자 오류 교정을 시도하였다. 실험 결과 자질 설계 등의 복잡한 작업 없이 워드 임베딩 만을 사용하여 해당 단어들에 대해 F1-measure 91.43 ~ 97.27%의 성능을 보였다.

• Proceedings of the Korean Environmental Sciences Society Conference
• /
• 2001.05a
• /
• pp.236-237
• /
• 2001

In this study, experiments were carried out to measure the variations in the corrosion potential and current density of polarization curves with amorphous $Fe_{70}Cr_5Si_{10}B_{15}$ and $Co_{70}Cr_5Si_{10}B_{15}$ alloy ribbon. The results were particularly examined to identify the influences of corrosion potential including various conditions such as hydrochloric acid, temperature, salt, pH, and oxygen. The optimum conditions were established with variations including temperature, salt, pH, oxygen, corrosion rate, and resistance of corrosion potential. The mass tranfer coefficient(${\alpha}$) value was determined with the Tafel's slope for the anodic dissolution based on the polarization effect with optimum conditions. The second anodic current density peak and maximum passive current density were designated as the critical corrosion sensitivity($I_{r}/I_{f}$).

• Annual Conference on Human and Language Technology
• /
• 2010.10a
• /
• pp.186-190
• /
• 2010

본 논문은 쇼핑몰의 이용 후기 중 광고성 리뷰를 분류해 내는 방법을 제안한다. 여기서 광고성 리뷰는 주로 업체에서 작성하는 것으로 리뷰 안에 광고 내용이 포함되어 있다. 국외 연구 중에는 드물게 오피니언 스팸 문서의 분류 연구가 진행되고 있지만 한국어 상품평으로부터 광고성 리뷰를 분류하는 연구는 아직 이루어지지 않고 있다. 본 논문에서는 Naive Bayes Classifier를 활용하여 광고성 리뷰를 분류하였다. 이때 확률 계산을 위해 사용된 특징 단어는 POS-Tagging+Bigram, POS-Tagging+Unigram, Bigram을 사용하여 추출하였다. 실험 결과는 POS-Tagging+Bigram 방법을 이용하였을 때 광고성 리뷰의 F-Measure가 80.35%로 정확도 높았다.

• Proceedings of the KSR Conference
• /
• 2008.11b
• /
• pp.1420-1425
• /
• 2008

This study is to examine the sound transmission loss of a railway vehicle floor. To this end, a semi-reverberation room was constructed. The semi-reverberation room was made of a railway vehicle floor between the sound radiating chamber and the sound receiving chamber. To block the sound, the wall was made of acryl, urethane foam, wood, and glass fiber. The test followed the KS F 2808 standard, and a typical reverberation room was used to verify the performance of the semi-reverberation room. As a result, comparison of the measurements showed that the test results of the semi-reverberation room had the same tendency as those of the reverberation room. Consequently it was possible to measure the sound transmission loss of railway vehicle structures using the semi-reverberation room.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2012.04a
• /
• pp.329-332
• /
• 2012

국내 택배시장 규모는 매출 3조원 이상, 물량 13 억 상자 이상을 처리하고 있다. 2000년 6천억원에서 불과 10년 사이에 500% 이상 확대되었다. 그에 반해 소비자들의 불만 역시 증가하였다. 따라서 현재의 수작업 VOC 분류 방식으로는 적정한 대응에 한계가 있을 수 밖에 없다. 이 논문에서는 효율적인 택배불만 처리를 위해서 불만의 종류와 정도를 기계학습을 이용하여 자동분류 하는 과정 및 결과를 기술한다. 약 93,000건의 VOC(voice of customer)를 대상으로 학습 데이터를 구축하고 여러 자질 선택 기법을 비교하였으며, 기존의 다양한 문서 자동 분류 방법들을 적용해 보았다. 실험결과 지지벡터기계가 가장 좋은 성능을 보였고, 각각의 F-measure 값은 불만의 정도는 83.1%, 불만의 종류는 75.9% 로 측정되었다.

• Healthcare Informatics Research
• /
• v.24 no.4
• /
• pp.376-380
• /
• 2018

Objectives: This research presents the design and development of a software architecture using natural language processing tools and the use of an ontology of knowledge as a knowledge base. Methods: The software extracts, manages and represents the knowledge of a text in natural language. A corpus of more than 200 medical domain documents from the general medicine and palliative care areas was validated, demonstrating relevant knowledge elements for physicians. Results: Indicators for precision, recall and F-measure were applied. An ontology was created called the knowledge elements of the medical domain to manipulate patient information, which can be read or accessed from any other software platform. Conclusions: The developed software architecture extracts the medical knowledge of the clinical histories of patients from two different corpora. The architecture was validated using the metrics of information extraction systems.

• International Journal of Computer Science & Network Security
• /
• v.21 no.4
• /
• pp.214-222
• /
• 2021

Recently, the growth of e-commerce in Saudi Arabia has been exponential, bringing new remarkable challenges. A naive approach for product matching and categorization is needed to help consumers choose the right store to purchase a product. This paper presents a machine learning approach for product matching that combines deep learning techniques with standard artificial neural networks (ANNs). Existing methods focused on product matching, whereas our model compares products based on unstructured descriptions. We evaluated our electronics dataset model from three business-to-consumer (B2C) online stores by putting the match products collectively in one dataset. The performance evaluation based on k-mean classifier prediction from three real-world online stores demonstrates that the proposed algorithm outperforms the benchmarked approach by 80% on average F1-measure.

• PNF and Movement
• /
• v.20 no.2
• /
• pp.235-241
• /
• 2022

Purpose: The purpose of this study was to identify the effects of low intensity blood flow restriction training (LBFR) on the central nervous system of healthy adults. Methods: Ten healthy right-handed adults (eight males and two females, mean age of 28.6 ± 2.87 years) were selected as study subjects. Functional magnetic resonance imaging (fMRI) was conducted to measure brain activation (BA) following LBFR and non-LBFR. The primary motor area, premotor area, and supplementary motor area, which are closely related to exercise, were set as the regions of interest. Results: The BA recorded during the LBFR condition was 931.7 ± 302.44 voxel, and the BA recorded during the non-LBFR condition was 1,510.9 ± 353.47 voxel. Conclusion: BA was lower during LBFR than during non-LBFR.

• Journal of Information Science Theory and Practice
• /
• v.11 no.1
• /
• pp.79-88
• /
• 2023

This study aimed to develop an ontology for Thai country songs by using the seven steps of an ontology development process. Hozo-Ontology Editor software and Ontology Application Management Framework were tools used in this study. Nine classes of ontology were identified: song, singer, emotion, author, language used, language type, song style, original, and content, and it was found that the song class had a relationship with all of the other classes. The developed ontology was evaluated by seeking opinions from experts in the field of Thai country songs, who agreed that the ontology was highly effective. Additionally, the evaluation employed the knowledge retrieval concept, and the precision, recall, and overall effectiveness were measured, with a precision of 92.59%, a recall of 86.21%, and an overall effectiveness (F-measure) of 89.28%. These results indicate that the developed ontology is highly effective in describing the scope of knowledge of Thai country songs.