• 제목/요약/키워드: dual language

검색결과 83건 처리시간 0.028초

Document Classification Methodology Using Autoencoder-based Keywords Embedding

  • Seobin Yoon;Namgyu Kim
    • 한국컴퓨터정보학회논문지
    • /
    • 제28권9호
    • /
    • pp.35-46
    • /
    • 2023
  • 본 연구에서는 문서 분류기의 정확도를 높이기 위해 문맥 정보와 키워드 정보를 모두 사용하는 이중 접근(Dual Approach) 방법론을 제안한다. 우선 문맥 정보는 다양한 자연어 이해 작업(Task)에서 뛰어난 성능을 나타내고 있는 사전학습언어모델인 Google의 BERT를 사용하여 추출한다. 구체적으로 한국어 말뭉치를 사전학습한 KoBERT를 사용하여 문맥 정보를 CLS 토큰 형태로 추출한다. 다음으로 키워드 정보는 문서별 키워드 집합을 Autoencoder의 잠재 벡터를 통해 하나의 벡터 값으로 생성하여 사용한다. 제안 방법을 국가과학기술정보서비스(NTIS)의 국가 R&D 과제 문서 중 보건 의료에 해당하는 40,130건의 문서에 적용하여 실험을 수행한 결과, 제안 방법이 문서 정보 또는 단어 정보만을 활용하여 문서 분류를 진행하는 기존 방법들에 비해 정확도 측면에서 우수한 성능을 나타냄을 확인하였다.

미셸 푸코와 건축의 근대성(I): - 말과 사물, 말과 건축 - (Michel Foucault and Modern Architecture(I) - Words and Things, Words and Architecture -)

  • 배형민
    • 건축역사연구
    • /
    • 제7권3호
    • /
    • pp.87-105
    • /
    • 1998
  • Surveying the literature of architecture since the nineteenth century, one can identify two dominant but problematic attitudes, among several, that pursue the task of defining what modern architecture is and should be. The first is the search for meaning and the second is the pursuit of form. This study, following Michel Foucault, asserts that the dual formation of meaning and form is a historical product of modernity and belies architecture's uncritical dependence on language since the nineteenth century. This study is a critique and historical analysis of this pernicious reliance, and constitutes a first step towards thinking of alternative relations between 'words and architecture' in the modern world. In reconstructing this problematic, the paper has called on Foucault's seminal The Order of Things. The study follows his construction of the Renaissance, the Classical and the Modern episteme, and in brief fashion, reconstructs the relation between language and architecture in each episteme. In analysing the Modern, the study focuses on Hegel's Lectures on Aesthetics. Hegel placed architecture in a genre hierarchy within which architecture, because of its material basis, was fundamentally limited in its ability to express the Spirit. For Hegel it was, among the arts, poetic language, and beyond art, the language of philosophy, through which the Absolute Spirit could be atttained. Much of post-nineteenth century architecture has remained within the shadow of Hegel, where architecture's materiality is perceived to be a burden, and in order to secure its relevance in modern society, architecture was deemed to pursue the role of language. As the most recent and sophisticated example of architecture's pursuit of form, the paper analyses the work of Peter Eisenman. Though Eisenman's theoretical writings are replete with post-Hegelian rhetoric, his architecture remains dependent upon the model of language, albeit a structuralist one. The paper concludes that ultimately, the pursuit of meaning and form is unable to face the crucial issue of value in modernity. While the former decides to easily what it is, the latter evades the issue itself. The second installment of this ongoing study will pursue a third possibility alluded to by Foucault, where language remains silent, pointing only to its 'ponderous' material existence.

  • PDF

Enhancement of a language model using two separate corpora of distinct characteristics

  • 조세형;정태선
    • 한국지능시스템학회논문지
    • /
    • 제14권3호
    • /
    • pp.357-362
    • /
    • 2004
  • 언어 모델은 음성 인식이나 필기체 문자 인식 등에서 다음 단어를 예측함으로써 인식률을 높이게 된다. 그러나 언어 모델은 그 도메인에 따라 모두 다르며 충분한 분량의 말뭉치를 수집하는 것이 거의 불가능하다. 본 논문에서는 N그램 방식의 언어모델을 구축함에 있어서 크기가 제한적인 말뭉치의 한계를 극복하기 위하여 두개의 말뭉치, 즉 소규모의 구어체 말뭉치와 대규모의 문어체 말뭉치의 통계를 이용하는 방법을 제시한다. 이 이론을 검증하기 위하여 수십만 단어 규모의 방송용 말뭉치에 수백만 이상의 신문 말뭉치를 결합하여 방송 스크립트에 대한 퍼플렉시티를 30% 향상시킨 결과를 획득하였다.

다문화시대 이중언어교육을 위한 교육연극 (Process Drama for Bilingual Education in Multiculturalism)

  • 한규용
    • 한국연극학
    • /
    • 제40호
    • /
    • pp.451-502
    • /
    • 2010
  • The thesis is to look for the method of applying process drama for bilingual education in multicultural society. This purpose can be resolved in three steps: first, to understand the purposes and functions of education changing to multicultural society; second, to survey the need of bilingual education conformed to its purposes and functions; last, to understand the theoretical background of process drama and seek the method of utilizing it to bilingual education. Recently, the Korean society is also changing to multicultural society with international marriages and immigrant workers. But they have difficulties in communicating in Korean, and their children have even the same. Under the basis of multiculturalism wanting peace and mutual respect between cultural groups, they have to be educated in equality. This is why the bilingual education is needed in our society. Process drama, a widely used term for educational drama in Britain, claims its effectiveness for L1, L2 and bilingual education, for the principal medium in drama is language and what drama delivers is related to meaningful human experiences. It is also supported on script theory of linguistic therapy. Kase-Polisini classifies the developmental process of education drama into 1) planning, 2) playing and 3) evaluation. This process can be applied to educational drama for language as well. Rather, educational drama applying to bilingual education has even more availability with the varieties of group organization and subject matter. Conclusively speaking, the group of dual language schools can be organized into 1) mother-tongued group, 2) mother-tongued and L2 group, 3) multilingual group, or 4) L3 group in educational drama activity, as the class generally consist of the mixed multilingual children. And the subject matter can deal with 1) the Korean society and culture, 2) the society and culture of the purposed language country, or 3) the society and culture of the third nations.

Putting Images into Second Language: Do They Survive in the Written Drafts?

  • Huh, Myung-Hye
    • 영어영문학
    • /
    • 제56권6호
    • /
    • pp.1255-1279
    • /
    • 2010
  • Much has already been learned about what goes on in the minds of second language writers as they compose, yet, oddly enough, until recently little in the L2 research literature has addressed writing and mental imagery together. However, images and imaging (visual thinking) play a crucial role in perception (the basis of mental imagery), in turn, affecting language, thinking, and writing. Many theorists of mental imagery also agree that more than just language accounts for how we think and that imagery is at least as crucial as language. All of these demands, to be sure, are compounded for EFL students, which is why I investigate EFL students' writing process, focusing on the use of mental imagery and its relationship to the writing. First I speculate upon some ways that imagery influences EFL students' composing processes and products. Next, I want to explore how and whether the images in a writer's mind can be shaped effectively into a linear piece of written English in one's writing. I studied two university undergraduate EFL students, L and J. They had fairly advanced levels of English proficiency and exhibited high level of writing ability, as measured by TOEFL iBT Test. Each student wrote two comparison and contrast essays: one written under specified time limitations and the other written without the pressure of time. In order to investigate whether the amount of time in itself causes differences within an individual in imagery ability, the students were placed under strict time constraints for Topic 1. But for Topic 2, they were encouraged to take as much time as necessary to complete this essay. Immediately after completing their essays, I conducted face-to-face retrospective interviews with students to prompt them for information about the role of imagery as they write. Both L and J have spent more time on their second (untimed) essays. Without time constraint, they produced longer texts on untimed essay (149 vs. 170; 186 vs 284 words). However, despite a relatively long period of time spent writing an essay, these students neither described their images nor detailed them in their essays. Although their mental imagery generated an explosion of ideas for their writings, most visual thinking must merely be a means toward an end-pictures that writers spent in purchasing the right words or ideas.

디지탈/아날로그 겸용 이동통신 단말기를 위한 오디오/데이타 프로세서의 실시간 구현에 관한 연구 (A Study of Real-Time Implementation of Audio/Data Processor for Digital/Analog Dual mode Mobile Phone)

  • 변경진;김종재;한기천;유하영;차진종;김경수
    • 한국음향학회지
    • /
    • 제16권2호
    • /
    • pp.80-88
    • /
    • 1997
  • 본 연구는 현재 디지탈 방식의 이동통신에서 사용되는 디지탈/아날로그 겸용 단말기에서 아날로그 방식을 지원하기 위한 오디오/데이타 프로세서를 ETRI DSP를 이용하여 실시간 구현하는 것에 대한 것이다. 오디오/데이타 프로세서는 단말기가 아날로그 방식으로 동작할 때 광대역 데이타 처리, 오디오신호 처리 및 demodulation, data rate conversion 기능을 수행한다. 이와같은 기능은 어셈블리 언어로 프로그램되어 디지탈 방식에서 사용되는 보코더 프로그램과 함께 ETRI DSP에 탑재되었다. 즉 하나의 하드웨어를 이용하여 디지탈 방식의 보코더와 아날로그 방식의 오디오/데이타 프로세서를 함께 구현 함으로써 하드웨어의 효율성을 극대화 하여 기존의 아날로그 전용의 단말기와의 경쟁력을 가질 수 있도록 하였다.

  • PDF

One-board micom을 이용한 정밀 온도 제어 시스템 (A precision temperature control system using one-board micom)

  • 주해호;조덕현
    • 제어로봇시스템학회:학술대회논문집
    • /
    • 제어로봇시스템학회 1988년도 한국자동제어학술회의논문집(국내학술편); 한국전력공사연수원, 서울; 21-22 Oct. 1988
    • /
    • pp.457-461
    • /
    • 1988
  • In this study an one-board micom controlled precision temperature control system has been developed. The digital temperature control system is consisted of an one-board micom as digital controller, a 12-bit A/D and D/A converter, a power amplifier, a NTC thermister, a preamplfier and a heat chamber. An operating control program for the control system was written in Z80 machine language. A Dual-PID predictor control algorithm was proposed. Experments were conducted with different sampling time and limitted error value. As a result, the temperature in a heat chamber can be well controlled within +- 0.2 .deg.C when the sampling time was applied to 10 sec and the limitted error value +- 0.5 .deg.C under the dual-PID predictor control algorithm. By means of one-board micom overall system has been reduced in size and volume, thus the system becomes compact and less expensive.

  • PDF

지그비 기반 센서 네트워크를 이용한 침수감지용 광 이중센서 개발 (Development of optical dual-sensors for submersion monitoring using zigbee-based wireless sensor networks)

  • 계광현;김형표;손경락
    • 센서학회지
    • /
    • 제19권3호
    • /
    • pp.184-190
    • /
    • 2010
  • In this paper, a remote submersion warning system based on multi-mode optical fiber(MMF) sensors and a wireless sensor network(WSN) are proposed. To improve the reliability and stability of the sensors, the dual optical fiber sensors combined to the optical coupler are demonstrated. A slave zigbee as a wireless sensor module was used as a platform to monitor and record the signal from the MMF sensors and then transmits these information to a master zigbee wirelessly. The monitoring system running the $LabVIEW^{(R)}$ software was connected to the internet to support the short message service(SMS) through extensible markup language(XML) web service. No matter where the managers are, they can always receive the real-time remote-monitoring data for safety check.

포스트모던 그림책에서의 역설과 아이러니 - 앤서니 브라운의 『돼지책』을 중심으로 (Paradox and Irony in Postmodern Picturebook - With Focus on Piggybook by Anthony Browne)

  • 나선희
    • 비교문화연구
    • /
    • 제38권
    • /
    • pp.65-87
    • /
    • 2015
  • This study has examined, first of all, one of unique characteristics of picturebook, so-called dual audience. To be more exact, this paper aimed at exploring the flexibility of picturebooks that allows both of children and adult reader to enjoy them. For this, we analysed Piggybook by Anthony Browne that so far has been read as a traditional picturebook with a single distinctive narrative rather than as a postmodern picturebook with multiple narratives, centering on paradox and irony. The results of this study were as fallows: Piggybook has the potential of multiple readings and meanings for a variety of audiences. Above all, this picturebook raises not only the question about the gender roles but also the one about representation of reality through the language.

뇌종양 분할을 위한 3D 이중 융합 주의 네트워크 (3D Dual-Fusion Attention Network for Brain Tumor Segmentation)

  • ;;;김수형
    • 한국정보처리학회:학술대회논문집
    • /
    • 한국정보처리학회 2023년도 춘계학술발표대회
    • /
    • pp.496-498
    • /
    • 2023
  • Brain tumor segmentation problem has challenges in the tumor diversity of location, imbalance, and morphology. Attention mechanisms have recently been used widely to tackle medical segmentation problems efficiently by focusing on essential regions. In contrast, the fusion approaches enhance performance by merging mutual benefits from many models. In this study, we proposed a 3D dual fusion attention network to combine the advantages of fusion approaches and attention mechanisms by residual self-attention and local blocks. Compared to fusion approaches and related works, our proposed method has shown promising results on the BraTS 2018 dataset.