• Annual Conference on Human and Language Technology
• /
• 1993.10a
• /
• pp.15-24
• /
• 1993

한국어는 구문요소의 문법기능이 표면구조상에 명시되는 구문특성을 갖고 있다. 이러한 특성은 한국어의 문법체계가 feature중심으로 전개되고 있음을 의미한다. 한국어에서의 feature 특성과 이진 결합 관계를 중심으로 하는 chart parsing 알고리즘을 제시하고 한국어 chart parser을 구현하였다.

• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.236-238
• /
• 2019

신경망 기계 번역(Neural Machine Translation)은 주로 지도 학습(Supervised learning)을 이용한 End-to-end 방식의 연구가 이루어지고 있다. 그러나 지도 학습 방법은 데이터가 부족한 경우에는 낮은 성능을 보이기 때문에 BERT와 같은 대량의 단일 언어 데이터로 사전학습(Pre-training)을 한 후에 미세조정(Finetuning)을 하는 Transfer learning 방법이 자연어 처리 분야에서 주로 연구되고 있다. 최근에 발표된 MASS 모델은 언어 생성 작업을 위한 사전학습 방법을 통해 기계 번역과 문서 요약에서 높은 성능을 보였다. 본 논문에서는 영어-한국어 기계 번역 성능 향상을 위해 MASS 모델을 신경망 기계 번역에 적용하였다. 실험 결과 MASS 모델을 이용한 영어-한국어 기계 번역 모델의 성능이 기존 모델들보다 좋은 성능을 보였다.

• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.629-632
• /
• 2021

기계 번역(machine translation)은 자연 언어로 된 텍스트를 다른 언어로 자동 번역 하는 기술로, 최근에는 주로 신경망 기계 번역(Neural Machine Translation) 모델에 대한 연구가 진행되었다. 신경망 기계 번역은 일반적으로 자기회귀(autoregressive) 모델을 이용하며 기계 번역에서 좋은 성능을 보이지만, 병렬화할 수 없어 디코딩 속도가 느린 문제가 있다. 비자기회귀(non-autoregressive) 모델은 단어를 독립적으로 생성하며 병렬 계산이 가능해 자기회귀 모델에 비해 디코딩 속도가 상당히 빠른 장점이 있지만, 멀티모달리티(multimodality) 문제가 발생할 수 있다. 본 논문에서는 단어 정렬(word alignment)을 이용한 비자기회귀 신경망 기계 번역 모델을 제안하고, 제안한 모델을 한국어-영어 기계 번역에 적용하여 단어 정렬 정보가 어순이 다른 언어 간의 번역 성능 개선과 멀티모달리티 문제를 완화하는 데 도움이 됨을 보인다.

• Annual Conference on Human and Language Technology
• /
• 1991.10a
• /
• pp.44-57
• /
• 1991

In the transfer machine translation, transfer dictionary decides the complexity of the transfer phase and the quality of translation according to the types and precision of informations supplied in the dictionary. Using the phrasal level translated informations within the human readable dictionary, human being translates a source sentence correctly and naturally. In this paper, we propose the verb transfer dictionary in which the various informations are constructed so the machine readable format that the Korean-to-English machine translation system can utilize them. In the proposed dictionary, we first provide the criterions by which an appropriate target verb is selected in phrase-to-phrase translations without an additional semantic analysis in transfer phase. Second, we provide the concrete sentence structure of a target verb so that we can resolve the expressive gaps between two languages and reduce the complexity of the various structure transfer in word-to-word translation.

• Annual Conference on Human and Language Technology
• /
• 1996.10a
• /
• pp.479-483
• /
• 1996

본 논문은 구운 분석의 검색 영역 축소를 통한 구문 분석기의 성능 향상을 목적으로 connectionism을 이용한 부분 구문 인식기의 설계와 구현을 기술한다. 본 부분 구문 인식기는 형태소 분석된 문장으로부터 명사-주어부와 술어부를 인식함으로써 전체 검색 영역을 여러 부분으로 나누어 구문 분석문제를 축소시키는 것을 목적으로 하고 있다. Connectionist 모델은 입력층과 출력층으로 구성된 개선된 퍼셉트론 구조이며, 입/출력층 사이의 노드들을, 입력층 사이의 노드들을 연결하는 연결 강도(weight)가 존재한다. 명사-주어부 및 술어부 구문 태그를 connectionist 모델에 적용하며, 학습 알고리즘으로는 개선된 백프로퍼게이션 학습 알고리즘을 사용한다. 부분 구문 인식 실험은 112개 문장의 학습 코퍼스와 46개 문장의 실험 코퍼스에 대하여 85.7%와 80.4%의 정확한 명사-주어부 및 술어부 인식을, 94.6%와 95.7%의 명사-주어부와 술어부 사이의 올바른 경계 인식을 보여준다.

• Journal of Environmental Health Sciences
• /
• v.41 no.2
• /
• pp.102-115
• /
• 2015

Objectives: A learning organization was designed and implemented on the basis of the selection criteria and essential elements of knowledge translation theory. Methods: The learning organization was designed on the basis of biosafety harmonization criteria and risk management strategy and was implemented as the learning organization for biosafety management by the National Institute of Health, Korea Centers for Disease Control & Prevention. The effect of knowledge translation in the research institutions by evidence-based policy was verified. Results: The result of applying the knowledge translation theory involving all stakeholders showed a positive reaction in establishing and implementing biosafety management strategy and embodied risk assessment criteria and evoked sympathy with the necessity of learning and using of expert knowledge about risk assessment and risk management. All stakeholders initiated voluntarily action toward new human-network construction and communication between similar organizations. The learning organization's capability expanded the base of knowledge translation. Conclusion: These results showed that a learning organization could enhance the autonomous safety management system by diffusion of knowledge translation.

• Genomics & Informatics
• /
• v.18 no.2
• /
• pp.23.1-23.6
• /
• 2020

The Human Phenotype Ontology (HPO) is the de facto standard ontology to describe human phenotypes in detail, and it is actively used, particularly in the field of rare disease diagnoses. For clinicians who are not fluent in English, the HPO has been translated into many languages, and there have been four initiatives to develop Japanese translations. At the Biomedical Linked Annotation Hackathon 6 (BLAH6), a rule-based approach was attempted to determine the preferable Japanese translation for each HPO term among the candidates developed by the four approaches. The relationship between the HPO and Mammalian Phenotype translations was also investigated, with the eventual goal of harmonizing the two translations to facilitate phenotype-based comparisons of species in Japanese through cross-species phenotype matching. In order to deal with the increase in the number of HPO terms and the need for manual curation, it would be useful to have a dictionary containing word-by-word correspondences and fixed translation phrases for English word order. These considerations seem applicable to HPO localization into other languages.

• BMB Reports
• /
• v.35 no.5
• /
• pp.513-517
• /
• 2002

We describe a new method for identifying the sequences that signal the start of translation, and the boundaries between exons and introns (donor and acceptor sites) in human mRNA. According to the mandatory keyword, ORGANISM, and feature key, CDS, a large set of standard data for each signal site was extracted from the ASCII flat file, gbpri.seq, in the GenBank release 108.0. This was used to generate the scoring matrices, which summarize the sequence information for each signal site. The scoring matrices take into account the independent nucleotide frequencies between adjacent bases in each position within the signal site regions, and the relative weight on each nucleotide in proportion to their probabilities in the known signal sites. Using a scoring scheme that is based on the nucleotide scoring matrices, the method has great sensitivity and specificity when used to locate signals in uncharacterized human genomic DNA. These matrices are especially effective at distinguishing true and false sites.

• Nutrition Research and Practice
• /
• v.8 no.4
• /
• pp.347-351
• /
• 2014

Inflammation is one mechanism through which cancer is initiated and progresses, and is implicated in the etiology of other conditions that affect cancer risk and prognosis, such as type 2 diabetes, cardiovascular disease, and visceral obesity. Emerging human evidence, primarily epidemiological, suggests that walnuts impact risk of these chronic diseases via inflammation. The published literature documents associations between walnut consumption and reduced risk of cancer, and mortality from cancer, diabetes, and cardiovascular disease, particularly within the context of the Mediterranean Diet. While encouraging, follow-up in human intervention trials is needed to better elucidate any potential cancer prevention effect of walnuts, per se. In humans, the far-reaching positive effects of a plant-based diet that includes walnuts may be the most critical message for the public. Indeed, appropriate translation of nutrition research is essential for facilitating healthful consumer dietary behavior. This paper will explore the translation and application of human evidence regarding connections with cancer and biomarkers of inflammation to the development of dietary guidance for the public and individualized dietary advice. Strategies for encouraging dietary patterns that may reduce cancer risk will be explored.

• Journal of the Korea Convergence Society
• /
• v.12 no.5
• /
• pp.1-7
• /
• 2021

In the latest trend of machine translation research, the model is pretrained through a large mono lingual corpus and then finetuned with a parallel corpus. Although many studies tend to increase the amount of data used in the pretraining stage, it is hard to say that the amount of data must be increased to improve machine translation performance. In this study, through an experiment based on the mBART model using parallel corpus filtering, we propose that high quality data can yield better machine translation performance, even utilizing smaller amount of data. We propose that it is important to consider the quality of data rather than the amount of data, and it can be used as a guideline for building a training corpus.