• 기호학연구
• /
• no.56
• /
• pp.137-158
• /
• 2018

In 1999, the French semiotician Jacques Fontanille published his work on discourse semiotics as an attempt to overcome the limitations of structuralism. Discourse semiotics has expanded the realm of semiotics by taking a phenomenological approach that examines the structure of consciousness. Phenomenology pays special attention to the operation of human consciousness, which, instead of staying put, moves from one place to another-unbeknownst to its agent. This paper examines "Bulls" through a phenomenological approach as the short story depicts the stream of consciousness that unfolds within the inner world of Pau, the young male protagonist. The rhetorical approach to literary analysis works at the level of the speech act and directly relates to speech act theory. In rhetorics, figures and tropes are dominated by speech acts. Whereas conventional rhetorical analysis lacks a kinetic perspective as it revolves around flat semantic inspections, the discourse semiotics approach applied in this paper allows for an indepth multidimensional analysis where the analyses of ${\acute{e}}nonc{\acute{e}}$ and ${\acute{e}}nonciation$ complement each other. Presenting a peasant movement to the reader through the eyes of the young male protagonist Pau, "Bulls" is considered a significant literary feat in the history of Korea's New Literature as the work is highly regarded for its impressive depiction of a communal movement. This paper reinterprets and reevaluates the story under a new light, mainly through a rhetorical approach and a phenomenological approach that hinges on discourse semiotics.

• Korean Journal of Cognitive Science
• /
• v.18 no.2
• /
• pp.179-191
• /
• 2007

In a goal-oriented dialogue, spoken' intentions can be represented by domain actions that consist of pairs of a speech art and a concept sequence. The domain action prediction of user's utterance is useful to correct some errors that occur in a speech recognition process, and the domain action prediction of system's utterance is useful to generate flexible responses. In this paper, we propose a model to predict a domain action of the next utterance using a neural network. The proposed model predicts the next domain action by using a dialogue history vector and a current domain action as inputs of the neural network. In the experiment, the proposed model showed the precision of 80.02% in speech act prediction and the precision of 82.09% in concept sequence prediction.

• Journal of the Korean Society of Propulsion Engineers
• /
• v.2 no.2
• /
• pp.1-13
• /
• 1998

When driven with a short (less than 30$\mu\textrm{s}$) low-energy pulse, the semiconductor bridge(SCB) produces a hot plasma that ignites explosive. The shape of plasma was observed using ultra high speed camera, the generation and the duration time of plasma were estimated by analyzing the ultra high speed camera image. The more energy supplied, the sooner the formation of the plasma was, and the size of the plasma was increased in proportion. The voltage variation of the bridge was measured and analyzed by comparing with the ultra high speed camera image.

• The Journal of the Korea Contents Association
• /
• v.8 no.7
• /
• pp.135-145
• /
• 2008

This paper tried to analyze Starcraft's multiplay by using especially Greimas's theory of discourse. Greimas's theory of semiotics is known as a useful tool of analyzing non-verbal text as well as verbal text. The discourse of Starcraft is continuously made by subject and anti-subject. This characteristic of discourse- not given completely made- is the prominent difference from discourses made from other media. The subject and anti-subject, as actively crossing the position of enunciator and enunciatee each other, dynamically make and complete the discourse.

• Annual Conference on Human and Language Technology
• /
• 1998.10c
• /
• pp.365-369
• /
• 1998

개념기반 한국어 대화체 분석 시스템에서 어려운 점으로 대두될 수 있는 것 중의 하나가 대화체 파싱에서 과다한 탐색공간의 생성이다. 과다한 탐색공간의 생성은 대화체 발화문으로부터 불필요한 탐색공간을 제거하는 메커니즘의 결여 때문이다. 따라서 본 논문에서는 이러한 문제점을 해결하고자 개념에 기반 되어서 작성된 문법을 통해서 얻어진 동사정보를 구성하여 단일 최상위 레벨 개념들로 분리하고 이를 가장 최소 개수의 최상위 레벨 개념으로 제한해서 제한된 개념으로 대화체 토큰열을 전사시키는 방법을 제시하였다. 그 결과 기존 탐색공간의 40%정도의 탐색공간을 제약할 수 있었다.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1998.08a
• /
• pp.409-412
• /
• 1998

미국의 카네기 메론 대학과 일본의 ATR 및 한국의 전자통신연구원 등이 가입한 CSTAR 에서는 99년 국제간 음성언어번역 시스템 데모를 위해 IF를 이용하여 데이터를 주고 받기로 합의하였다. IF는 크게는 인터넷을 통해 다른 나라의 음성언어번역 시스템과 연결하여 데이터를 주고 받는데 사용되고, 작게는 음성언어 번역 시스템 내의 해석 시스템과 생성 시스템 사이에 데이터를 주고 받는데 사용된다. IF는 중간언어 표현의 한 가지 방법으로 간단하면서도 단순한 표현으로 특정 영역 내에 나타나는 이미를 표현할 수 있도록 정의되었다. 대상으로 하는 영역은 여행 안내로 호텔 예약, 비행기 예약, 여행지 안내 및예약 등을 포함하고 있다. IF의 가장 큰 특징은 표현방법의 단순화에 있다. 즉, 의미를 가장 잘 나타낼 수 있는 표현을 골라, IF를 정의하여 언어 종속적인 요소를 가능한 배제하였다. IF 태깅은 발화에 대해 적절한 IF를 붙여 주는 일로 태깅을 수행하는 사람은 IF 태깅 요령에 따라 태깅을 수행하여야 한다. 현재 ETRI에서는 200대화 이상의 한국어 데이터에 대해 IF 태깅을 완료하였으며 해석 시스템과 생성 시스템 개발을 계속하고 있다.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.153-159
• /
• 2022

본 논문에서는 인공지능이 생성하는 일상 대화의 품질 향상을 위해 상식 추론을 정의하고 설문을 통해 정량적, 정성적 분석을 진행하였다. 정량적 평가에서는 주어진 문장이 에게 학습시키기에 적합한가'라는 수용성 판단을 요청한 질문에서 40대 이상의 연령이 20, 30대와 유의미한 차이를 보였다. 정성적 평가에서는 '보편적 사실 여부'를 AI 발화 기준의 주요한 지표로 보았다. 이어서 '챗봇' 대화의 품질에 대한 설문을 실시했다. 이를 통해 일상 대화를 사용한 챗봇의 대화 품질을 높이기 위해서는 먼저, 질문의 요구에 적절한 정보와 공감을 제공해야 하고 두 번째로 공감의 정도가 챗봇의 특성에 맞는 응답이어야 하며 세 번째로 대화의 차례에 따라 담화의 규칙을 지키면서 대화가 진행되어야 한다는 결론을 얻을 수 있었다. 이 세 가지 요건이 통합적으로 적용된 담화 설계를 통해 완전히 인공지능스러운 대화가 가능할 것으로 여겨진다.

• Journal of the Korea Convergence Society
• /
• v.13 no.1
• /
• pp.13-21
• /
• 2022

Recent deep learning-based research shows excellent performance in most natural language processing (NLP) fields with pre-trained language models. In particular, the auto-encoder-based language model proves its excellent performance and usefulness in various fields of Korean language understanding. However, the decoder-based Korean generative model even suffers from generating simple sentences. Also, there is few detailed research and data for the field of conversation where generative models are most commonly utilized. Therefore, this paper constructs multi-turn dialogue data for a Korean generative model. In addition, we compare and analyze the performance by improving the dialogue ability of the generative model through transfer learning. In addition, we propose a method of supplementing the insufficient dialogue generation ability of the model by extracting recommended response candidates from external knowledge information through a retrival model.

• Applied Chemistry for Engineering
• /
• v.9 no.2
• /
• pp.165-171
• /
• 1998

The $Y_1Ba_2Cu_3O_{7-x}$(123) superconductor powders were prepared by pyrophoric synthesis method(PSM) using $Y_2O_3$(99.9%), $BaCO_3$(99.9%), and CuO(99.9%) powders. The phase formation and reaction kinetics of 123 superconductor manufactured with powders prepared in various pHs of pyrophoric synthetic solution have been studied through the experiments at various heat treatment temperatures and times. Inductively coupled plasma(ICP) spectroscopy and scanning electron microscopy(SEM) measurements were performed to examine the composition and morphology of the sample. X-ray diffraction(XRD) analysis was done to determine phase formation and conversion ratio of Y-Ba-Cu-O systems. The 123 powder prepared at pH 7(${\pm}0.3$) yields the best result in terms of purity, homogeneity, and reactivity. The activation energies(${\Delta}E_a$) of 123 phase formation were found to be 191 kJ/mol and 230kJ/mol in solid state reaction method and pyrophoric synthesis method, respectively.

• Journal of Digital Convergence
• /
• v.11 no.4
• /
• pp.403-408
• /
• 2013

The configuration analysis for the image appeared in the storyboard as the 1st stage to change the scenario to the image expressed the visual phenomenon image starting from the literalism. The simultaneity of description on the interpretative level of the narrative description may describe various information multiple simultaneously, the relationship has been presented as a lot of these information has been weaved as one episode. The incident time of narrative description as the scene described mainly objectively and illustratively has been used usefully to present the accurate information for characters or incidents. The time series of narrative description appeared as one episode by expressing various spaces or images under the continuity of flow according to the time. The collision image seen from the angle of internal ignition played a role to create the meaning colliding into each other or interconnecting symbolic effects appeared as respectively fragmented image, the continuity of space played a role of expressing the different symbol when the image of completely different space is connected to one context, the continuity of time as well makes a different symbolism by interpreting symbols appeared as each images under the flow of time. These results show that the narrative description may be expressed in accordance with the narrative structure from the viewpoints of narrative description, but the internal ignition may be interpreted through the cultural and periodical background widely known in those years based on the experience and information capability the audiences are keeping not by any descriptive structure. Images appeared in the internal ignition appeared by colliding into each other or amplifying mutually no related scenes and revealing the overall symbolism by reinterpreting again.