• ETRI Journal
• /
• v.43 no.6
• /
• pp.1049-1057
• /
• 2021

Automatic spacing in Korean is used to correct spacing units in a given input sentence. The demand for automatic spacing has been increasing owing to frequent incorrect spacing in recent media, such as the Internet and mobile networks. Therefore, herein, we propose a transformer encoder that reads a sentence bidirectionally and can be pretrained using an out-of-task corpus. Notably, our model exhibited the highest character accuracy (98.42%) among the existing automatic spacing models for Korean. We experimentally validated the effectiveness of bidirectional encoding and pretraining for automatic spacing in Korean. Moreover, we conclude that pretraining is more important than fine-tuning and data size.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2000.10a
• /
• pp.261-264
• /
• 2000

지금까지의 한영 번역 방식은 규칙 기반 방식이 주를 이루었지만 현재 패턴을 이용한 번역 방식이 활발히 연구되고 있다. 그러나 패턴 기반 방식은 그 적용성(Coverage)에 대한 치명적인 단점을 지닌다. 따라서 본 논문에서는 한국어 패턴을 어절 단위의 일반 문틀과 동사구를 중심으로 하는 용언중심의 문틀로 나누어 각 패턴들에 대한 적용성 및 실현성을 조사한다. 실험은 기존의 형태소 분석기를 이용하여 방송 자막 문장 351,806 문장을 대상으로 자동으로 문틀을 구축하여 4,995 문장의 테스트 데이터에 대한 적용성 검사를 실시하였다. 즉 본 논문에서는 방송 자막 문장을 대상으로 한영번역을 위한 일반 문틀 및 용언 중심의 문틀 방식의 적용성을 조사하여 문틀 기반 방식의 실현성을 평가하고 앞으로의 한영 번역 시스템 개발 방향을 제시한다.

• Journal of the Korea Convergence Society
• /
• v.12 no.12
• /
• pp.39-47
• /
• 2021

The construction of high-quality input features through effective segmentation is essential for increasing the sentence comprehension of a language model. Improving the quality of them directly affects the performance of the downstream task. This paper comparatively studies the segmentation that effectively reflects the linguistic characteristics of Korean regarding word and sentence classification. The segmentation types are defined in four categories: eojeol, morpheme, syllable and subchar, and pre-training is carried out using the RoBERTa model structure. By dividing tasks into a sentence group and a word group, we analyze the tendency within a group and the difference between the groups. By the model with subchar-level segmentation showing higher performance than other strategies by maximal NSMC: +0.62%, KorNLI: +2.38%, KorSTS: +2.41% in sentence classification, and the model with syllable-level showing higher performance at maximum NER: +0.7%, SRL: +0.61% in word classification, the experimental results confirm the effectiveness of those schemes.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.31B no.3
• /
• pp.14-23
• /
• 1994

A natural language processing system usually has the demerit that its processing time is relatively long. If an interactive system makes its user kept waiting long, it can't be said to be practical. In this paper, the on-line natural language parser in which its processing coincides with the sentence's inputting is designed. Since the greater part of morpholgical and syntatic semantic analysis is already performed during the keyboard input, user can get a prompt response. Moreover, the Korean parser is implemented in multitasking environment, and it is compared with an off-line parser. The on-line parser can be considered to be efficient for its real time processing.

• Phonetics and Speech Sciences
• /
• v.13 no.1
• /
• pp.53-64
• /
• 2021

According to Demand and Capacity Model (DCM), external and internal environments influence the disfluency of children who stutter (CWS). This study investigated the effects of simultaneous changes in motoric and linguistic demands on CWS and children who do not stutter (CWNS). Participants were 4-6 years old CWS and CWNS. A sentence imitation task with changes in speech rate, utterance length, and sentence complexity was used to examine their effects on children's disfluency. When the utterance length changed, CWS showed more disfluency regardless of utterance length and as the speech rate changed, CWS showed more disfluency at fast speech rate than CWNS. When the utterance length and speech rate changed, at fast speech rate, CWS showed more disfluency in both utterances than CWNS. When sentence complexity changed, CWS showed more disfluency than CWNS in complex sentences. Changes in linguistic elements such as speech rate, utterance length, and sentence complexity affect disfluency in CWS, especially when they were exposed to faster, longer, and more complex sentences. This indicates that CWS are vulnerable to fast and complex speech motor control and language processing ability than CWNS. Thus, this study suggests that parents and therapists consider both the speech rate and the utterance length when talking with CWS.

• Korean Journal of Cognitive Science
• /
• v.17 no.2
• /
• pp.103-118
• /
• 2006

The main purpose of this paper is to investigate the effects of discourse topic and morpho-syntactic verbal information on the resolution of null pronouns in the Korean narrative text within the framework of the construction-integration theory (Kintsch, 1988, Singer & Kintsch, 2001, Graesser, Gernsbacher, & Goldman. 2003). For the purpose of this paper, two conditions were designed: an explicit condition with both a consistently maintained discourse topic and the person-specific verb modals on one hand, and a neutral condition with no discourse topic or morpho-syntactic information provided, on the other. We measured the reading tines far the target sentence containing a null pronoun and the question response times for finding an antecedent, and the accuracy rates for finding an antecedent. During the experiments each passage was presented at a tine on a computer-controlled display. Each new sentence was presented on the screen at the moment the participant pressed the button on the computer keyboard. Main findings indicate that processing is facilitated by macro-structure (topicality) in conjunction with micro-structure (morpho-syntax) in pronoun interpretation. It is speculated that global processing alone may not be able to determine which potential antecedent is to be focused unless aided by lexical information. It is argued that the results largely support the resonance-based model, but not the minimalist hypothesis.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.7
• /
• pp.291-298
• /
• 2022

In this paper, we present a method for dependency parsing with chunking in Korean. Dependency parsing is a task of determining a governor of every word in a sentence. In general, we used to determine the syntactic governor in Korean and should transform the syntactic structure into semantic structure for further processing like semantic analysis in natural language processing. There is a notorious problem to determine whether syntactic or semantic governor. For example, the syntactic governor of the word "먹고 (eat)" in the sentence "밥을 먹고 싶다 (would like to eat)" is "싶다 (would like to)", which is an auxiliary verb and therefore can not be a semantic governor. In order to mitigate this somewhat, we propose a Korean dependency parsing after chunking, which is a process of segmenting a sentence into constituents. A constituent is a word or a group of words that function as a single unit within a dependency structure and is called a chunk in this paper. Compared to traditional dependency parsing, there are some advantage of the proposed method: (1) The number of input units in parsing can be reduced and then the parsing speed could be faster. (2) The effectiveness of parsing can be improved by considering the relation between two head words in chunks. Through experiments for Sejong dependency corpus, we have shown that the USA and LAS of the proposed method are 86.48% and 84.56%, respectively and the number of input units is reduced by about 22%p.

• Annual Conference on Human and Language Technology
• /
• 2002.10e
• /
• pp.325-331
• /
• 2002

정보추출 시스템의 목적은 관심의 대상이 되는 특정 정보를 선택적으로 찾아내 제시하는데 있다. 따라서 도메인 정보에 의존적인 방법으로 정보추출이 이루어질 수밖에 없고, 이에 따른 도메인 정보 구축의 부담이 컸다. 이러한 부담을 줄이기 위해 본 연구에서는 특정 주제영역과 관련한 문서로부터 자동으로 이벤트 문장을 추출하는 시스템을 제안한다. 이벤트 문장이란, 특정도메인에서 다루어지는 이벤트의 구체적인 내용을 포함하고 있는 문장이다. 이러한 문장을 추출함으로써 기본적인 수준의 정보추출 요구를 만족시킬 수 있을 뿐만 아니라, 주출된 이벤트 문장을 도메인 정보 구축에 활용할 수 있을 것이다. 본 연구에서는 동사, 명사, 명사구, 및 3W 자질을 이용하여 문장추출의 성능을 최대화하기 위한 방안을 제안하고, 세 개의 평가 도메인을 대상으로 실험을 수행하였다. 실험 결과, when 및 where 자질과 동사, 명사. 명사구의 가중치를 이용하여 문장 가중치를 계산함으로써 최적의 이벤트 문장추출 성능을 얻을 수 있음을 알 수 있었다.

• Proceedings of the Korean Society for Cognitive Science Conference
• /
• 2006.06a
• /
• pp.67-71
• /
• 2006

현재까지 제안된 자동 띄어쓰기 교정 모델들은 그 중의 대다수가 입력 문장에서 공백을 제거한 후에 교정 작업을 수행한다. 이러한 교정 방식은 입력 문장의 띄어쓰기가 잘 되어 있는 경우에 입력 문장보다 좋지 못한 교정 문장을 생성하는 경우가 있다. 본 논문에서는 이러한 문제점을 해결하기 위하여 입력 문장의 띄어쓰기를 고려한 자동 띄어쓰기 교정모델을 제안한다. 이 모델은 입력 문장의 음절단위 띄어쓰기 오류가 5%일 때 약 8%의 성능 향상을 보였으며, 10%의 오류가 존재할 때 약 5%의 성능 향상을 보였다.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.5
• /
• pp.1317-1329
• /
• 2000

The advanced techniques for video processing and computer graphics enables a sign language education system to appear. the system is capable of showing a sign language motion for an arbitrary sentence using the captured video clips of sign language words. In this paper, a method is suggested which generates the frames between the last frame of a word and the first frame of its following word in order to animate hand motion. In our method, we find hand locations and angles which are required for in between frame generation, capture and store the hand images at those locations and angles. The inbetween frames generation is simply a task of finding a sequence of hand angles and locations. Our method is computationally simple and requires a relatively small amount of disk space. However, our experiments show that inbetween frames for the presentation at about 15fps (frame per second) are achieved so tat the smooth animation of hand motion is possible. Our method improves on previous works in which computation cost is relativey high or unnecessary images are generated.