• 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.

• Proceedings of the Korean Society for Language and Information Conference
• /
• 1996.02a
• /
• pp.383-392
• /
• 1996

• Proceedings of the Korean Society for Language and Information Conference
• /
• 2007.11a
• /
• pp.375-384
• /
• 2007

This paper describes a method for automatic acquisition of wide-coverage treebank-based deep linguistic resources for Japanese, as part of a project on treebank-based induction of multilingual resources in the framework of Lexical-Functional Grammar (LFG). We automatically annotate LFG f-structure functional equations (i.e. labelled dependencies) to the Kyoto Text Corpus version 4.0 (KTC4) (Kurohashi and Nagao 1997) and the output of of Kurohashi-Nagao Parser (KNP) (Kurohashi and Nagao 1998), a dependency parser for Japanese. The original KTC4 and KNP provide unlabelled dependencies. Our method also includes zero pronoun identification. The performance of the f-structure annotation algorithm with zero-pronoun identification for KTC4 is evaluated against a manually-corrected Gold Standard of 500 sentences randomly chosen from KTC4 and results in a pred-only dependency f-score of 94.72%. The parsing experiments on KNP output yield a pred-only dependency f-score of 82.08%.

• Proceedings of the Korean Information Science Society Conference
• /
• 2003.04c
• /
• pp.513-515
• /
• 2003

본 논문에서는 코퍼스를 이용한 확률 의존문법 자동 생성 기술을 다룬다. 의존문법 생성을 위해 구성성분의 기능어들 간의 의존관계를 학습했던 기존 연구와는 달리. 한국어 구성성분은 내용어와 기능어의 결함 형태로 구성되고 임의 구성성룬 기능어와 임의 구성성분 내용어간의 의존관계가 의미가 있다는 사실을 반영한 의존문법 학습방법을 제안한다. KAIST의 트리 부착 코퍼스 31,086문장에서 추출한 30,600문장의 Tagged Corpus을 가지고 학습한 결과 초기문법을 64%까지 줄인 1.101 개의 의존문법을 획득했고. 실험문장 486문장을 Parsing한 결과 73.81%의 Parsing 정확도를 보였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2001.10a
• /
• pp.507-510
• /
• 2001

본 논문에서는 코퍼스를 이용한 확률 의존문법 자동 생성 기술을 다룬다. 특히 의존 문법 생성을 위해 확률 재추정 알고리즘을 의존문법생성에 맞도록 변형하여 학습하였으며 정확한 문법 생성 및 회귀데이터(Data Sparseness)문제 해결을 위해서 구성요소의 대표 지배소들 간의 의존관계 만을 학습했던 기존 연구와는 달리 구성요소 내부의 의존관계까지 학습하는 방법을 제안한다. KAIST 의 트리 부착 코퍼스 31,086 문장에서 추출한 25,000 문장의 Tagged Corpus 을 가지고 한국어 확률 의존 문법 학습을 시도 하였다. 그 결과 초기문법을 10.97% 에서 23.73% 까지 줄인 2,349 개의 정확한 문법을 얻을 수 있었다. 문법의 정확성을 실험 하기 위해 350 개의 실험문장을 Parsing 한 결과 69.61%의 파싱 정확도를 보였다. 이로서 구성요소 내부의 의존관계 학습으로 얻어진 의존문법이 더 정확했으며, 회귀데이터 문제 또한 극복할 수 있음을 알 수 있었다.

• Journal of the Korea Society of Computer and Information
• /
• v.11 no.2 s.40
• /
• pp.319-327
• /
• 2006

Higher grade elementary school students who have superior cognitive abilities need education of basic principles of computer or programming rather than computer in education. In this study, all the errors occurring while elementary school students wrote and executed programs were collected. in the method of predicting and dealing with possible-to-occur problems on programming education of the higher grades (4th, 5th and 6th grades) during their optional special activities or during talent aptitude activities after school, classified by type and analyzed. If the errors analyzed are put to practical use, optimal programming curriculums could be written and such curriculums could be a great contribution to induction of learning effect and interest on teaching learning. It was found by analyzing the errors collected for this study that the most of elementary school students during programming felt difficulties in simple errors by poor use of software and in simple coding by poor use of reserved words in English. In the next, students occurred errors by difficulties in understanding grammar. It was exposed that these error types were the opposite phenomena to those analyzed by commercial software developing companies, however, it is predicted that if teaching learning is setting improved, the same phenomena could be found desirably.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.1
• /
• pp.135-140
• /
• 2023

VR animation allows you to see 360-degree screen production that cannot be seen in existing animation production while being located in the space within the animation. <Goodbye Mr Octopus>, a VR animation produced in 2020, was selected at the 77th Venice Film Festival as an immersive short film. This is the story of an adolescent girl, Stella, who is celebrating her 16th birthday, and her conflict with her strict father is resolved through a letter from her mother. It is a narrative composed of a total of 11 scenes, and in each scene, new directing elements of VR video grammar, such as gaze induction, time flow, and space conversion directing, were analyzed. Gaze-inducing directing minimized the inconvenience of 360-degree gaze, and time and space conversion directing was analyzed as an effect of increasing the audience's immersion according to narrative events.