• Annual Conference on Human and Language Technology
• /
• 2006.10e
• /
• pp.148-155
• /
• 2006

한국어 구문 분석에서 가장 어려운 작업들 중에 하나는 종속절의 의존관계 파악이다. 본 논문에서는 이를 해결하기 위해서 종속절의 의존관계를 걸을 구성하는 서술어부(동사와 어미)의 관련 정보의 유무에 따라 의존관계가 성립한다고 가정했다. 즉 각각의 절들의 서술부의 관련 정보의 유무로 보고, 이진 분류 문제로 이 문제를 해결하였다. 사용한 자질은 정적 자질(static feature)와 동적 자질(dynamic feature)를 구성되어 있다. 정적 자질은 동사와 어미에서 표면적인 어휘 정보이고 이는 단어, POS 테그 및 위치 정보들이다. 동적 자질은 문장에서 절이 가지는 문법적인 형태를 의미하고, 이를 추출하기 위해 간단한 규칙을 만들고 이를 바탕으로 CKY 차트 파서를 통하여 추출하였다. 기계학습 방법으로는 이진 분류 문제에서 널리 사용되는 SVM을 사용하였다. 실험 결과 어휘 정보들 중에서 어미의 정보만 사용하였을 경우는 64.4%의 정확도를 보였고 문법적인 정보인 동적 자질을 사용한 경우는 73.5%로 어휘 정보만을 사용한 경우 보다 9.1%의 성능 향상됨을 보였다

• Annual Conference on Human and Language Technology
• /
• 2010.10a
• /
• pp.73-77
• /
• 2010

본 논문은 의존 구조 분석기를 통합하기 위해 입력 문장의 단어 특성을 활용하는 단어 단위 분배기를 제안한다. 본 모델은 기존의 문장 수준 분배기와는 달리 입력 문장의 단어 특성에 따라 가장 적절한 의존 구조 분석기를 선택하고, 선택된 의존 구조 분석기의 결과를 최종 결과로 사용한다. 기존의 문장 단위 분배기보다 단어 수준의 풍부한 특질을 활용할 수 있다는 장점과 큰 크기의 코퍼스를 사용할 수 있다는 장점이 있다. 총 6개 언어의 LAS를 측정했는데, MALT 보다는 평균 1.98%, MST 보다는 0.54%의 성능 향상이 있었다.

• The KIPS Transactions:PartB
• /
• v.9B no.6
• /
• pp.791-798
• /
• 2002

In this thesis. we present a new method for inducing a probabilistic dependency grammar (PDG) from text corpus. As words in Korean are composed of a set of more basic morphemes, there exist various dependency relations in a word. So, if the induction process does not take into account of these in-word dependency relations, the accuracy of the resulting grammar nay be poor. In comparison with previous PDG induction methods. the main difference of the proposed method lies in the fact that the method takes into account in-word dependency relations as well as inter-word dependency relations. To access the performance of the proposed method, we conducted an experiment using a manually-tagged corpus of 25,000 sentences which is complied by Korean Advanced Institute of Science and Technology (KAIST). The grammar induction produced 2,349 dependency rules. The parser with these dependency rules shoved 69.77% accuracy in terms of the number of correct dependency relations relative to the total number dependency relations for best-1 parse trees of sample sentences. The result shows that taking into account in-word dependency relations in the course of grammar induction results in a more accurate dependency grammar.

• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.181-184
• /
• 2019

상호참조해결은 주어진 문서에서 상호참조해결 대상이 되는 멘션(mention)을 식별하고, 같은 개체(entity)를 의미하는 멘션을 찾아 그룹화하는 자연어처리 태스크이다. 한국어 상호참조해결에서는 멘션 탐지와 상호참조해결을 동시에 진행하는 end-to-end 모델과 포인터 네트워크 모델을 이용한 방법이 연구되었다. 구글에서 공개한 BERT 모델은 자연어처리 태스크에 적용되어 많은 성능 향상을 보였다. 본 논문에서는 한국어 상호참조해결을 위한 BERT 기반 end-to-end 신경망 모델을 제안하고, 한국어 데이터로 사전 학습된 KorBERT를 이용하고, 한국어의 구조적, 의미적 특징을 반영하기 위하여 의존구문분석 자질과 개체명 자질을 적용한다. 실험 결과, ETRI 질의응답 도메인 상호참조해결 데이터 셋에서 CoNLL F1 (DEV) 71.00%, (TEST) 69.01%의 성능을 보여 기존 연구들에 비하여 높은 성능을 보였다.

• Annual Conference on Human and Language Technology
• /
• 1993.10a
• /
• pp.559-570
• /
• 1993

In this article we present a rapid development of a Korean to English translation system, by the help of general English generator, PENMAN. PENMAN is an English sentence generation system, of which input language is a language specially devised for sentence generation, named Sentence Planning Language(SPL). The language SPL has various features that are necessary for generating sentences, covering both syntactic and semantic features. In this development we integrated a Korean language parser based on dependency grammar and the English sentence generator PENMAN, bridging two systems through a converting module, which converts dependency structures produced by Korean parser into SPL for PENMAN.

• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.161-177
• /
• 2019

In this paper, we study the performance improvement of the answer extraction in Question-Answering system by using sentence dependency parsing result. The Question-Answering (QA) system consists of query analysis, which is a method of analyzing the user's query, and answer extraction, which is a method to extract appropriate answers in the document. And various studies have been conducted on two methods. In order to improve the performance of answer extraction, it is necessary to accurately reflect the grammatical information of sentences. In Korean, because word order structure is free and omission of sentence components is frequent, dependency parsing is a good way to analyze Korean syntax. Therefore, in this study, we improved the performance of the answer extraction by adding the features generated by dependency parsing analysis to the inputs of the answer extraction model (Bidirectional LSTM-CRF). The process of generating the dependency graph embedding consists of the steps of generating the dependency graph from the dependency parsing result and learning the embedding of the graph. In this study, we compared the performance of the answer extraction model when inputting basic word features generated without the dependency parsing and the performance of the model when inputting the addition of the Eojeol tag feature and dependency graph embedding feature. Since dependency parsing is performed on a basic unit of an Eojeol, which is a component of sentences separated by a space, the tag information of the Eojeol can be obtained as a result of the dependency parsing. The Eojeol tag feature means the tag information of the Eojeol. The process of generating the dependency graph embedding consists of the steps of generating the dependency graph from the dependency parsing result and learning the embedding of the graph. From the dependency parsing result, a graph is generated from the Eojeol to the node, the dependency between the Eojeol to the edge, and the Eojeol tag to the node label. In this process, an undirected graph is generated or a directed graph is generated according to whether or not the dependency relation direction is considered. To obtain the embedding of the graph, we used Graph2Vec, which is a method of finding the embedding of the graph by the subgraphs constituting a graph. We can specify the maximum path length between nodes in the process of finding subgraphs of a graph. If the maximum path length between nodes is 1, graph embedding is generated only by direct dependency between Eojeol, and graph embedding is generated including indirect dependencies as the maximum path length between nodes becomes larger. In the experiment, the maximum path length between nodes is adjusted differently from 1 to 3 depending on whether direction of dependency is considered or not, and the performance of answer extraction is measured. Experimental results show that both Eojeol tag feature and dependency graph embedding feature improve the performance of answer extraction. In particular, considering the direction of the dependency relation and extracting the dependency graph generated with the maximum path length of 1 in the subgraph extraction process in Graph2Vec as the input of the model, the highest answer extraction performance was shown. As a result of these experiments, we concluded that it is better to take into account the direction of dependence and to consider only the direct connection rather than the indirect dependence between the words. The significance of this study is as follows. First, we improved the performance of answer extraction by adding features using dependency parsing results, taking into account the characteristics of Korean, which is free of word order structure and omission of sentence components. Second, we generated feature of dependency parsing result by learning - based graph embedding method without defining the pattern of dependency between Eojeol. Future research directions are as follows. In this study, the features generated as a result of the dependency parsing are applied only to the answer extraction model in order to grasp the meaning. However, in the future, if the performance is confirmed by applying the features to various natural language processing models such as sentiment analysis or name entity recognition, the validity of the features can be verified more accurately.

• Annual Conference on Human and Language Technology
• /
• 1992.10a
• /
• pp.73-81
• /
• 1992

한 언어에 대한 포괄적인 언어정보 데이타베이스의 구축에 있어서는 수집된 텍스트에 대한 상세한 문법정보의 주석이 일차적 작업 대상이 된다. 이는 통사적 정보가 단순히 구문 분석상의 문제들을 해결하기 위한 정보를 제공해주는 것일 뿐 아니라 형태소 해석 및 문장 의미의 파악등 자연언어 이해시스템 전반의 성능을 향상시키는 데에 중요한 물을 차지하기 때문이다. 각개 단어의 문법적 기능에 대한 주석은 사전적 정의에 따른다면 "품사"로 표현할 수 있을 것이다. 그런데 품사는 각개 단어가 지니는 고유한 어휘의미적 정보이기보다는 구문구조에 의존적인 양상을 보인다. 이는 사전에 따라서 각개 단어에 대한 품사 정보가 달리 나타나는 점에서도 간취할 수 있는데, 한편으로 한국어 언어정보 데이타베이스 구축을 위한 문법적 주석에 있어서는 기존 사전의 품사정보에만 의존할 수는 없다는 문제점이 제기된다. 따라서 각 어휘들의 구문정보(흑은 품사정보)를 어떻게 기술할 것인가가 해결되어야 하는 것이다. 본 연구에서는 일차적으로 각 어휘들의 문장 안에서의 기능을 바탕으로 한 주석체계를 설정하고 그에 따라서 약 12만개의 문장에 대한 일차적 형식화를 수작업으로 처리하였다. 이는 향후 자동적으로 문법적 주석이 가능하도록 해주는 시스템의 개발을 지원하기 위한 언어정보의 수집에 목적을 둔 것인데, 이를 통해서 기존 국어사전에서의 언어정보상의 미비점을 수정 보완할 몇 가지 근거를 마련할 수 있었다.

• Cross-Cultural Studies
• /
• v.53
• /
• pp.163-192
• /
• 2018

The purpose of this paper was to examine universal dependency UD application cases of Korean and Japanese with similar morphological characteristics. In addition, UD application and improvement methods of Korean were examined through comparative analysis. Korean and Japanese are very well developed due to their agglutinative characteristics. Therefore, there are many difficulties to apply UD which is built around English refraction. We examined the application of UPOS and DEPREL as components of UD with discussions. In UPOS, we looked at category problem related to narrative such as AUX, ADJ, and VERB, We examined how to handle units. In relation to the DEPREL annotation system, we discussed how to reflect syntactic problem from the basic unit annotation of syntax tags. We investigated problems of case and aux arising from the problem of setting dominant position from Korean and Japanese as the dominant language. We also investigated problems of annotation of parallel structure and setting of annotation basic unit. Among various relation annotation tags, case and aux are discussed because they show the most noticeable difference in distribution when comparing annotation tag application patterns with Korean. The case is related to both Korean and Japanese surveys. Aux is a secondary verb in Korean and an auxiliary verb in Japanese. As a result of examining specific annotation patterns, it was found that Japanese aux not only assigned auxiliary clauses, but also auxiliary elements to add the grammatical meaning to the verb and form corresponding to the end of Korean. In UD annotation of Japanese, the basic unit of morphological analysis is defined as a unit of basic syntactic annotation in Japanese UD annotation. Thus, when using information, it is necessary to consider how to use morphological analysis unit as information of dependency annotation in Korean.

• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.61-65
• /
• 2021

본 연구에서는 2018년에 공개된 Penn Korean Universal Dependency Treebank(이하 PKT-UD v2018) 데이터의 오류를 분석하고 이를 개정하여 새롭게 데이터셋(이하 PKT-UD v2020)을 구축하였다. PKT-UD v2018은 구구조 분석 방식으로 구축된 Penn Korean Treebank를 UD(Universal Dependencies)의 체계에 맞추어 자동적으로 변환한 후 보정하여 구축한 데이터이다. 본 연구에서는 이와 같은 자동 변환의 과정에서 발생한 오류를 바로 잡고, UD 체계를 최대한 활용하면서 한국어의 특성을 잘 살린 데이터셋을 구축할 수 있는 방법을 제안하였다.

• The KIPS Transactions:PartB
• /
• v.13B no.1 s.104
• /
• pp.63-70
• /
• 2006

In general, a corpus contains lots of linguistic information and is widely used in the field of natural language processing and computational linguistics. The creation of such the corpus, however, is an expensive, labor-intensive and time-consuming work. To alleviate this problem, annotation tools to build corpora with much linguistic information is indispensable. In this paper, we design and implement an annotation tool for establishing a Korean dependency tree-tagged corpus. The most ideal way is to fully automatically create the corpus without annotators' interventions, but as a matter of fact, it is impossible. The proposed tool is semi-automatic like most other annotation tools and is designed to edit errors, which are generated by basic analyzers like part-of-speech tagger and (partial) parser. We also design it to avoid repetitive works while editing the errors and to use it easily and friendly. Using the proposed annotation tool, 10,000 Korean sentences containing over 20 words are annotated with dependency structures. For 2 months, eight annotators have worked every 4 hours a day. We are confident that we can have accurate and consistent annotations as well as reduced labor and time.