• 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.

• The Journal of the Acoustical Society of Korea
• /
• v.15 no.4
• /
• pp.24-31
• /
• 1996

In this paper, we present a speech translation system for hotel reservation, KT_STS(Korea Telecom Speech Translation System). KT-STS is a speech-to-speech translation system which translates a spoken utterance in Korean into one in Japanese. The system has been designed around the task of hotel reservation(dialogues between a Korean customer and a hotel reservation de나 in Japan). It consists of a Korean speech recognition system, a Korean-to-Japanese machine translation system and a korean speech synthesis system. The Korean speech recognition system is an HMM(Hidden Markov model)-based speaker-independent, continuous speech recognizer which can recognize about 300 word vocabularies. Bigram language model is used as a forward language model and dependency grammar is used for a backward language model. For machine translation, we use dependency grammar and direct transfer method. And Korean speech synthesizer uses the demiphones as a synthesis unit and the method of periodic waveform analysis and reallocation. KT-STS runs in nearly real time on the SPARC20 workstation with one TMS320C30 DSP board. We have achieved the word recognition rate of 94. 68% and the sentence recognition rate of 82.42% after the speech recognition tests. On Korean-to-Japanese translation tests, we achieved translation success rate of 100%. We had an international joint experiment in which our system was connected with another system developed by KDD in Japan using the leased line.

• The Journal of the Convergence on Culture Technology
• /
• v.8 no.1
• /
• pp.239-244
• /
• 2022

Neural Machine Translation (NMT) is gradually breaking down the boundary between human and machine translation. We look at actual cases of human and machine translation and discuss why machine translation needs a human touch. In this paper, we raise three driving questions: Can humans be replaced by machines?; How human translators can remain successful in a NMT-driven world?; Is it possible to eliminate language barrier in the era of NMT and World Englishes? The answers to these questions are all negative. We suggest that machine translation is a useful tool with rapidity, accuracy, and low cost productivity. However, the machine translation is limited in the areas of culture, borrowing, ambiguity, new words and (national) dialects. The machines cannot imitate the emotional and intellectual abilities of human translators since machines are based on machine learning, while humans are on intuition. The machine translation will be a useful tool that does not cause moral problems when using methods such as back translation and human post-editing. To conclude, we propose the blended approach that machine translation cannot be completed without the touch of human translation.

• The Journal of the Korea Contents Association
• /
• v.21 no.12
• /
• pp.505-516
• /
• 2021

The purpose of this study is to develop sign language interpreting content in the medical setting that facilitates Deaf people's access to medical services in situations where professional and accurate medical sign language interpreting is insufficient. To achieve the purpose of this study, we conducted a literature review, individual interviews for Deaf people, on-site requirement surveys of sign language interpreters and sign language experts, and medical and sign language expert consultations. Based on this, we developed sign language interpreting content such as main care contextual scenarios, basic medical terms, and medical term descriptions. Through this study, we developed medical sign language content considering the situation and medical importance of Deaf people to promote expertise in the medial sign language area and developed a responsive website of sign language medical dictionary that effectively and efficiently delivers information to Deaf people and sign language interpreters; we realized the need and importance of sign language translation for Deaf people to be the main bodies.

• Journal of the Korea Convergence Society
• /
• v.9 no.6
• /
• pp.217-224
• /
• 2018

The aim of this study is to examine how satires in the English movie, are translated into Korean. Satire is a literary technique in which the writer expresses sarcasm or criticism by using humor or irony. As satire induces laughter by criticising someone or something in the form of humor, it is not easy to convey the same effect to audiences with different social, cultural and political backgrounds. Naturally, satire translation poses great challenge to translators. This study analyzed satirical humors in , based on 4 basic strategies discussed commonly in the previous studies on humor translation, which found out Source Text(ST) preservation prevails, that is to say literal translation. This analysis result is expected to be worth in drawing an effective strategy for satire translation in the convergence perspective of society, culture or politics in other countries.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.16 no.1
• /
• pp.124-129
• /
• 2006

This parer presents a method of ranking translation word selection for Korean verbs based on lexical knowledge contained in a bilingual Korean-English dictionary and WordNet that are easily obtainable knowledge resources. We focus on deciding which translation of the target word is the most appropriate using the measure of semantic relatedness through the 45 extended relations between possible translations of target word and some indicative clue words that play a role of predicate-arguments in source language text. In order to reduce the weight of application of possibly unwanted senses, we rank the possible word senses for each translation word by measuring semantic similarity between the translation word and its near synonyms. We report an average accuracy of $51\%$ with ten Korean ambiguous verbs. The evaluation suggests that our approach outperforms the default baseline performance and previous works.

• International journal of advanced smart convergence
• /
• v.10 no.2
• /
• pp.86-96
• /
• 2021

This cross-sectional survey research carried out with the inquisitive agenda on satisfaction of the translation outcomes as performed by human translation and (machine translation) post-editing. The survey group consisted of 166 Korean translators primarily working with the English, Chinese, and Japanese languages. They were asked to rate the satisfactory level with accuracy, fluency, idiomatic expression, and terminology in the Richter's scale of four. The result reveals that human translation is more satisfactory than post-editing with respect to accuracy, but it is uneasy to assert that accuracy is unsatisfactory in post-editing. On the other hand, the Korean translators are less satisfied with fluency, idiomatic expression, and terminology than accuracy. It can be assumed that although human translation is more satisfactory than post-editing, the accuracy of post-editing seems to be more acknowledged than fluency, idiomatic expression, and terminology, which lead the translators to take the accuracy of raw machine-translation products and to go on to improve the fluency, idiomatic expression, and terminology. Nevertheless, Korean translators believe Korean idiomatic expressions cannot be satisfactorily produced in post-editing, while fluency and terminology can be improved in post-editing.

• Proceedings of the IEEK Conference
• /
• 2003.07b
• /
• pp.819-822
• /
• 2003

This paper defines an algorithmic description language in cycle-accurate level which can be used to design hardware components. The proposed language is less complex and more flexible than VHDL language. In that the language includes C-like control flow descriptions and brief timing information(i.e. clock cycle boundaries) indicated by 'wait_edge' statements. We generate RTL VHDL codes from the descriptions. The proposed language requires only 10～30% of the # of lines to describe the same functionality compared with RTL VHDL.