• 정보처리학회논문지B
• /
• 제13B권1호
• /
• pp.63-70
• /
• 2006

말뭉치(corpus)는 많은 언어 정보를 포함하고 있으며, 언어처리 및 계산언어학 분야에서 다양한 용도로 사용되고 있다. 그러나 말뭉치에 언어 정보를 부착하는 데는 많은 시간과 인력이 소요된다. 이 문제를 완화시키기 위해서 말뭉치 구축 도구가 반드시 요구된다. 본 논문에서는 한국어 의존구조 부착을 위한 말뭉치 구축 도구의 설계 및 구현에 관해서 기술한다. 가장 이상적인 방법은 주석자가 전혀 개입하지 않고, 말뭉치를 구축하는 것이나 이것은 사실상 불가능하다. 따라서 대부분의 말뭉치 구축 도구는 반자동으로 구성되어 있으며, 본 논문에서 제안된 도구도 반자동이다. 제안된 도구는 언어 분석기의 분석 결과에 내포된 오류를 효과적으로 수정할 수 있고, 또한 가능한 한 반복적인 작업을 피할 수 있으며 쉽게 사용할 수 있도록 인터페이스를 설계하였다. 제안된 시스템을 이용해서 20어절 이상의 1만 문장에 의존구조를 부착해 보았다. 잘 훈련된 8명의 주석자들이 매일 4시간씩 2개월 동안 구축하였으며, 그 결과는 정확하고 일관성 있는 말뭉치를 구축할 수 있었으며, 작업 시간과 인력도 크게 줄일 수 있었다.

• Genomics & Informatics
• /
• 제18권2호
• /
• pp.20.1-20.4
• /
• 2020

Accessible negative results are relevant for researchers and clinicians not only to limit their search space but also to prevent the costly re-exploration of research hypotheses. However, most biomedical relation extraction datasets do not seek to distinguish between a false and a negative relation among two biomedical entities. Furthermore, datasets created using distant supervision techniques also have some false negative relations that constitute undocumented/ unknown relations (missing from a knowledge base). We propose to improve the distinction between these concepts, by revising a subset of the relations marked as false on the phenotype-gene relations corpus and give the first steps to automatically distinguish between the false (F), negative (N), and unknown (U) results. Our work resulted in a sample of 127 manually annotated FNU relations and a weighted-F1 of 0.5609 for their automatic distinction. This work was developed during the 6th Biomedical Linked Annotation Hackathon (BLAH6).

• 한국언어정보학회:학술대회논문집
• /
• 한국언어정보학회 2007년도 정기학술대회
• /
• pp.358-364
• /
• 2007

Syntax parsers can benefit from speakers' intuition about constituent structures indicated in the input string in the form of parentheses. Focusing on languages like Korean, whose orthographic convention requires more than one word to be written without spaces, we describe an algorithm for passing the bracketing information across the tagger to the probabilistic CFG parser, together with one for heightening (or penalizing, as the case may be) probabilities of putative constituents as they are suggested by the parser. It is shown that two or three constituents marked in the input suffice to guide the parser to the correct parse as the most likely one, even with sentences that are considered long.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2011년도 한국컴퓨터종합학술대회논문집 Vol.38 No.1(C)
• /
• pp.258-261
• /
• 2011

Living in the age of information people face problems related to information overload. Information is easy to produce, store and distribute through various communication channels, one of which is emails. With the appearance of the mobile devices, such as smart phones and tabs, people can have access to email inbox at any moment of time from everywhere. In this paper we present information extraction system with a specific goal of extracting meeting location from the announcement of seminar or conference. We apply a machine learning method (conditional random fields, CRF), train the system using annotated corpus of seminar and conference announcements and validate results by applying various extracted correction rules and patterns. Furthermore, we normalize extracted location, and reference using geo-coding databases, OpenStreetMap and Wikipedia resources to determine real geographical coordinates.

• 음성과학
• /
• 제8권1호
• /
• pp.77-91
• /
• 2001

This paper investigates the timing of Korean spoken in a news-reading speech style in order to improve the naturalness of durations used in Korean speech synthesis. Each segment in a corpus of 671 read sentences was annotated with 69 segmental and prosodic features so that the measured duration could be correlated with the context in which it occurred. A CART model based on the features showed a correlation coefficient of 0.79 with an RMSE (root mean squared prediction error) of 23 ms between actual and predicted durations in reserved test data. These results are comparable with recent published results in Korean and similar to results found in other languages. An analysis of the classification tree shows that phrasal structure has the greatest effect on the segment duration, followed by syllable structure and the manner features of surrounding segments. The place features of surrounding segments only have small effects. The model has application in Korean speech synthesis systems.

• 한국어학
• /
• 제81권
• /
• pp.245-265
• /
• 2018

The purpose of this study is to prepare the basis for a more objective evaluation of oral fluency by analyzing Korean native speaker's utterance. Traditionally, fluency evaluation tended to rely on the evaluators' experience and subjective idea. Therefore, there has been a need of setting the evaluation standard in numeric form that is easily measurable. In this study, I will analyze Korean native speaker's utterance in focus of pause. Total number of 875 pauses were extracted from the 21st Century Sejong Korean spoken corpus, and the elements before and after the pauses were annotated. From the analysis results, the pauses were divided between fluent pauses and influent pauses. If the length of fluent pauses do not exceed reasonable length of pause for native Korean speakers, there was no point reduction. On the other hand, if the influent pauses are made more frequently than the native Korean speakers, then it is subject to point reduction.

• Journal of information and communication convergence engineering
• /
• 제18권4호
• /
• pp.207-215
• /
• 2020

Part of speech (POS) tagging is an indispensable part of major NLP models. Its progress can be perceived on number of languages around the globe especially with respect to European languages. But considering Indian Languages, it has not got a major breakthrough due lack of supporting tools and resources. Particularly for Odia language it has not marked its dominancy yet. With a motive to make the language Odia fit into different NLP operations, this paper makes an attempt to develop a POS tagger for the said language on a HMM (Hidden Markov Model) platform. The tagger judiciously considers bigram HMM with dynamic Viterbi algorithm to give an output annotated text with maximum accuracy. The model is experimented on a corpus belonging to tourism domain accounting to a size of approximately 0.2 million tokens. With the proportion of training and testing as 3:1, the proposed model exhibits satisfactory result irrespective of limited training size.

• 정보처리학회논문지B
• /
• 제19B권1호
• /
• pp.43-52
• /
• 2012

의미 역 결정 (Semantic Role Labeling)은 문장의 각 요소들의 의미 관계를 파악하는 연구 분야로써 어휘 중의성 해소와 더불어 자연언어처리에서의 의미 분석에서 매우 중요한 위치를 차지하고 있다. 그러나 한국어의 경우에는 의미 역 결정에 필요한 언어 자원이 구축되지 못하여 연구의 진행이 매우 미진한 상황이다. 본 논문에서는 의미 역 결정에 필요한 언어 자원 중에서 가장 널리 사용되고 있는 PropBank의 한국어 버전의 구축을 위한 시작 단계로써 자동 술어-논항 분석기를 개발하였다. 자동 술어-논항 분석기는 크게 의미 어휘 사전과 자동 술어-논항 추출기로 구성된다. 의미 어휘 사전은 한국어 동사의 격틀 정보를 구축한 사전이며 자동 술어-논항 추출기는 구문 표지 부착된 말뭉치로부터 특정 술어와 관련있는 논항의 의미 부류를 결정하는 모듈이다. 본 논문에서 개발된 자동 술어-논항 분석기는 향후 한국어 PropBank의 구축을 용이하게 할 것이며, 궁극적으로는 한국어 의미 역 결정에 큰 역할을 할 것이다.

• 말소리와 음성과학
• /
• 제5권1호
• /
• pp.99-108
• /
• 2013

Previous laboratory studies have shown that prosodic structures are encoded in the modulations of phonetic patterns of speech including suprasegmental as well as segmental features. In particular, effects of prosodic context on duration and intensity of syllables and words have been widely reported. Drawing on prosodically annotated large-scale speech data from the Buckeye corpus of conversational speech of American English, the current study attempted to examine whether and how prosodic prominence and phrase boundary of everyday conversational speech, as determined by a large group of ordinary listeners, are related to the phonetic realization of duration and intensity. The results showed that the patterns of word durations and intensities are influenced by prosodic structure. Closer examinations revealed, however, that the effects of prosodic prominence are not the same as those of prosodic phrase boundary. With regard to intensity measures, the results revealed the systematic changes in the patterns of overall RMS intensity near prosodic phrase boundary but the prominence effects are restricted to the nucleus. In terms of duration measures, both prosodic prominence and phrase boundary are the most closely related to the lengthening of the nucleus. Yet, prosodic prominence is more closely related to the lengthening of the onset while phrase boundary lengthens the coda duration more. The findings from the current study suggest that the phonetic realizations of prosodic prominence are different from those of prosodic phrase boundary, and speakers signal different prosodic structures through deliberate modulations of the internal phonetic structure of words and listeners attend to such phonetic variations.

• 한국언어정보학회지:언어와정보
• /
• 제15권1호
• /
• pp.31-62
• /
• 2011

TimeML is a markup language for events and temporal expressions in natural language, proposed in Pustejovsky et al. (2003) and latter standardized as ISO-TimeML (ISO 24617-1:2009). In this paper, we propose the further specification of ISO-TimeML for the Korean language with the concrete and thorough examination of real world texts. Since Korean differs significantly from English, which is the first and almost only extensively tested language with TimeML, one continuously run into theoretical and practical difficulties in the application of TimeML to Korean. We focus on the discussion for the consistent and efficient application of TimeML: how to consistently apply TimeML in accordance with Korean specificity and what to be annotated and what not to be, i.e. which information is meaningful in the temporal interpretation of Korean text, for efficient application of TimeML.