• Educational Technology International
• /
• 제10권2호
• /
• pp.145-166
• /
• 2009

The purpose of this study was to investigate why advance organizers (AO) are effective in promoting comprehension and mental model formation in terms of cognitive load. Two experimental groups: a concept-map AO group and a key-word AO group and one control group were used. This study considered cognitive load in view of Baddeley's working memory model: central executive (CE), phonological loop (PL), and visuo-spatial sketch pad (VSSP). The present experiment directly examined cognitive load using dual task methodology. The results were as follows: central executive (CE) suppression task achievement for the concept map AO group was higher than the key word AO group and control group. Comprehension and mental model construction for the concept map AO group were higher than the other groups. These results indicated that the superiority of concept map AO owing to CE load decrement occurred with comprehension and mental model construction in learning. Thus, the available resources produced by CE load reduction may have been invested for comprehension and mental model construction of learning contents.

• ETRI Journal
• /
• 제41권3호
• /
• pp.371-382
• /
• 2019

Machine reading comprehension is the task of understanding a given context and finding the correct response in that context. A simple recurrent unit (SRU) is a model that solves the vanishing gradient problem in a recurrent neural network (RNN) using a neural gate, such as a gated recurrent unit (GRU) and long short-term memory (LSTM); moreover, it removes the previous hidden state from the input gate to improve the speed compared to GRU and LSTM. A self-matching network, used in R-Net, can have a similar effect to coreference resolution because the self-matching network can obtain context information of a similar meaning by calculating the attention weight for its own RNN sequence. In this paper, we construct a dataset for Korean machine reading comprehension and propose an $S^2-Net$ model that adds a self-matching layer to an encoder RNN using multilayer SRU. The experimental results show that the proposed $S^2-Net$ model has performance of single 68.82% EM and 81.25% F1, and ensemble 70.81% EM, 82.48% F1 in the Korean machine reading comprehension test dataset, and has single 71.30% EM and 80.37% F1 and ensemble 73.29% EM and 81.54% F1 performance in the SQuAD dev dataset.

• ETRI Journal
• /
• 제41권6호
• /
• pp.771-781
• /
• 2019

We propose the VS³-NET model to solve the task of question answering questions with machine-reading comprehension that searches for an appropriate answer in a given context. VS³-NET is a model that trains latent variables for each question using variational inferences based on a model of a simple recurrent unit-based sentences and self-matching networks. The types of questions vary, and the answers depend on the type of question. To perform efficient inference and learning, we introduce neural question-type models to approximate the prior and posterior distributions of the latent variables, and we use these approximated distributions to optimize a reparameterized variational lower bound. The context given in machine-reading comprehension usually comprises several sentences, leading to performance degradation caused by context length. Therefore, we model a hierarchical structure using sentence encoding, in which as the context becomes longer, the performance degrades. Experimental results show that the proposed VS³-NET model has an exact-match score of 76.8% and an F1 score of 84.5% on the SQuAD test set.

• 영어어문교육
• /
• 제15권4호
• /
• pp.345-364
• /
• 2009

This study attempts to provide a comprehensive framework for listening strategy use among university students in Korea in relation to TOEIC listening scores. In particular, this study tests whether motivation, based on the self-determination theory, mediates listening strategy use on listening comprehension (LC) process and whether reading comprehension ability moderates the use of listening strategy in LC achievement. One hundred seventy six freshmen students participated in the study during their first semester required English course. Self-report questionnaires were used to assess students' motivation and LC strategy use. The responses were statistically analyzed via the moderator and mediator model. The results indicate that internalized motivation mediates the use of listening strategy in LC achievement; however, reading comprehension skill doesn't affect students' use of listening strategies in relation to listening skill achievement. In other words, students who have internalized motivation were able to utilize listening strategies effectively in terms of achievement of the TOEIC listening skills. The findings of the current study offer in-depth understanding of the relationship among use of LC strategies, intrinsic motivation, and listening skill achievement shared by the mediator and moderator models.

• 음성과학
• /
• 제8권4호
• /
• pp.185-191
• /
• 2001

This study was designed to determine what components of spoken language have been relatively neglected in the teaching of listening comprehension in Korea and to suggest a model of listening process. Two types of tests were undertaken using spoken and written forms of English with secondary school teachers of English and college students. Findings: Hearing power has been generally neglected in the teaching of listening comprehension. Hearing power which can be thought as an active process is defined as an ability to transfer the sequence of discrete phonetic segments without word boundary into the sequence of words in phonemic representations by using both nonlinguistic factors and linguistic factors including perception rules based on phonetics and phonology. Vocabularies, hearing-speaking power, syntactic structures and idiomatic expressions are to be taught for spoken English. A model of listening process was suggested and discussed.

• 음성과학
• /
• 제10권4호
• /
• pp.241-254
• /
• 2003

The purpose of the present study was to investigate the sentence comprehension strategies used by Korean/English bilingual children when they listened to sentences of their first language, i.e., Korean. The framework of competition model was employed to analyze the influence of the second language, i.e., English, during comprehension of Korean sentences. The participants included 10 bilingual children (ages 7;4-13;0) and 20 Korean-speaking monolingual children(ages 5;7-6;10) with similar levels of development in Korean language as bilingual children. In an act-out procedure, the children were asked to determine the agent in sentences composed of two nouns and a verb with varying conditions of three cues (case-marker, animacy, and word-order). The results revealed that both groups of children used the case marker cues as the strongest cue among the three. The bilingual children relied on case-marker cues even more than the monolingual children. However, the bilingual children used animacy cues significantly less than the monolingual children. There were no significant differences between the groups in the use of word-order cues. The bilingual children appeared less effective in utilizing animacy cues in Korean sentence comprehension due to the backward transfer from English where the cue strength of animacy is very weak. The influence of the second language on the development of the first language in bilingual children was discussed.

• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제7권2호
• /
• pp.69-76
• /
• 2018

참조 표현이란 장면 영상 내의 특정 물체를 가리키는 자연어 문장들을 의미한다. 본 논문에서는 참조 표현 이해를 위한 새로운 심층 신경망 모델을 제안한다. 본 논문에서 제안하는 모델은 장면 영상 내 대상 물체의 영역을 찾아내기 위해, 참조 표현에서 언급하는 대상 물체뿐만 아니라 보조 물체, 그리고 대상 물체와 보조 물체 사이의 관계까지 풍부한 정보를 활용한다. 또한 제안 모델에서는 영상 내 각 후보 영역의 적합도 계산을 위해 물체 적합도와 관계 적합도를 참조 표현의 문장 구조에 따라 결합한다. 따라서, 본 모델은 크게 총 네 가지 서브 네트워크들로 구성된다: 언어 표현 네트워크(LRN), 물체 정합 네트워크(OMN), 관계 정합 네트워크(RMN), 그리고 가중 결합 네트워크(WCN). 본 논문에서는 세 가지 서로 다른 참조 표현 데이터집합들을 이용한 실험을 통해, 제안 모델이 현존 최고 수준의 참조 표현 이해 성능을 보인다는 것을 입증하였다.

• 한국멀티미디어학회논문지
• /
• 제23권2호
• /
• pp.351-360
• /
• 2020

If various unreasonable safety standards are preemptively and effectively readjusted, the risk of accidents can be reduced. In this paper, we proposed a machine reading comprehension-based safety standard Q&A system to secure supporting technology for effective search and analysis of safety standards for integrated and systematic management of safety standards. The proposed model finds documents related to safety standard questions in the various laws and regulations, and then divides these documents into provisions. Only those provisions that are likely to contain the answer to the question are selected, and then the BERT-based machine reading comprehension model is used to find answers to questions related to safety standards. When the proposed safety standard Q&A system is applied to KorQuAD dataset, the performance of EM 40.42% and F1 55.34% are shown.

• 인지과학
• /
• 제14권3호
• /
• pp.23-36
• /
• 2003

학습에 대한 이해-기반 접근에 따르면 새로운 정보는 기존의 배경지식과 통합되어 정신표상을 형성하며 이는 다른 새로운 정보를 결합하는데 사용된다고 가정한다. 지시문을 통한 학습상황에서 인간과 계산적 모형의 수행비교를 통해 이 접근법이 타당하다는 것을 보여주었다. 구성-통합 이론 (Kintsch, 1988; 1998)에 근거한 계산적 모형 (ADAPT-UNIX)은 사용자들이 UNIX 복합 명령문을 생성하는데 도움을 주기위해 제시된 지시문 학습에 높은 예측력을 보였다. 더불어, 제시된 지시문을 사용하여 올바른 복합명령문을 생성하는 과제수행도 실제 인간수행과 높은 유사성 보였다. 배경지식의 수준에 따라 지시문이 학습과 적용에 차별적인 영향을 미친다는 교육적 함의와 이해-기반 인지모델의 이론적 함의가 논의되었다.

• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제9권4호
• /
• pp.145-152
• /
• 2020

기계독해는(Machine reading comprehension) 사용자 질의와 관련된 문서를 기계가 이해한 후 정답을 추론하는 인공지능 자연어처리 태스크를 말하며, 이러한 기계독해는 챗봇과 같은 자동상담 서비스에 활용될 수 있다. 최근 자연어처리 분야에서 가장 높은 성능을 보이고 있는 BERT 언어모델은 대용량의 데이터를 pre-training 한 후에 각 자연어처리 태스크에 대해 fine-tuning하여 학습된 모델로 추론함으로써 문제를 해결하는 방식이다. 본 논문에서는 BERT기반 특허상담 기계독해 태스크를 위해 특허상담 데이터 셋을 구축하고 그 구축 방법을 소개하며, patent 코퍼스를 pre-training한 Patent-BERT 모델과 특허상담 모델학습에 적합한 언어처리 알고리즘을 추가함으로써 특허상담 기계독해 태스크의 성능을 향상시킬 수 있는 방안을 제안한다. 본 논문에서 제안한 방법을 사용하여 특허상담 질의에 대한 정답 결정에서 성능이 향상됨을 보였다.