• Journal of the Korea Society of Computer and Information
• /
• v.26 no.10
• /
• pp.157-165
• /
• 2021

Recently, studies using deep learning to analyze a large amount of text are being actively conducted. In particular, a pre-trained language model that applies the learning results of a large amount of text to the analysis of a specific domain text is attracting attention. Among various pre-trained language models, BERT(Bidirectional Encoder Representations from Transformers)-based model is the most widely used. Recently, research to improve the performance of analysis is being conducted through further pre-training using BERT's MLM(Masked Language Model). However, the traditional MLM has difficulties in clearly understands the meaning of sentences containing new words such as newly coined words. Therefore, in this study, we newly propose NTM(Newly coined words Target Masking), which performs masking only on new words. As a result of analyzing about 700,000 movie reviews of portal 'N' by applying the proposed methodology, it was confirmed that the proposed NTM showed superior performance in terms of accuracy of sensitivity analysis compared to the existing random masking.

• Proceedings of the Korean Society of Medical Physics Conference
• /
• 2002.09a
• /
• pp.439-441
• /
• 2002

The aim of this study is to reconstruct the 3D target volume from multiple projection images. It was assumed that we were already aware of the target position exactly, and all processes were performed in Target Coordinates whose origin was the center of the target. We used six projections: two projections were used to make a Reconstruction Box and four projections were for image acquisition. Reconstruction Box was made up of voxels of 3D matrix. Projection images were transformed into 3D volume in this virtual box using geometrical based back-projection method. Algorithm was applied to an ellipsoid model and horse-shoe shaped model. Projection images were created using C program language by geometrical method and reconstruction was also accomplished using C program language and Matlab(The Mathwork Inc., USA). For ellipsoid model, reconstructed volume was slightly overestimated but target shape and position was proved to be correct. For horse-shoe shaped model, reconstructed volume was somewhat different from original target model but there was a considerable improvement in target volume determination.

• Cross-Cultural Studies
• /
• v.19
• /
• pp.313-338
• /
• 2010

Attitudes toward target language culture and motivation have been known as important factors in second or foreign language learning. In this study, cadets' attitudes toward the target language community and culture were investigated to find out any relationships with English achievement. Integrative or instrumental motivation in foreign military context was investigated to determine which motivation was more positively related to English proficiency. Cadets' responses were shown highly positive and internally consistent in most cases. Military cadets' attitudes toward American culture were statistically significant and could predict their English achievement. Even though integrative motivation was statistically significant, correlation with English achievement was low. However, instrumental motivation was not significant. This result implies that attitudes toward culture is an important factor in foreign language learning and integrative-instrumental motivation dichotomy might not be enough to explain specific context like cadets' English learning situation at Korea Army Academy at Youngcheon.

• Proceedings of the Korean Society for Language and Information Conference
• /
• 1995.02a
• /
• pp.109-114
• /
• 1995

• English Language & Literature Teaching
• /
• no.1
• /
• pp.159-187
• /
• 1995

The purpose of this study is to investigate common errors made by Korean learners in English composition and to find out what is an effective and appropriate teaching method of English composition in Korea. For these purposes, 197 students on the third grade in high school were selected as the subjects of this research. The students were tested by way of the immediate translation of 31 simple Korean sentences into English which are supposed to be easy for those students to write without any difficulty. About 2 minutes were given for testing each sentence. The results are as follows : First. the whole sum of errors made by 197 students was 2,972 and these types of errors were classified into 13 categories by Duskova's grammatical method and James'. The errors with comparatively high frequency were prepositional errors(17.2%), verbal errors(15.4%), and the errors with low frequency were article errors(1.9%), to-infinitive errors. Second, when Korean students learn English as a target language, overgeneralization(33.6%) and reduction(17.5) influenced the learners much more greatly than language transfer(22.2) did. But the influence of language transfer including interference & overgeneralization(l5.2%) and interference & reduction(10.7%) was no less than 48.1%. The statistics shows that the learners have a tendency to analyze, systematize and regularize the target language when they start to learn a new language.

• Proceedings of the Korean Society for Language and Information Conference
• /
• 1995.02a
• /
• pp.29-38
• /
• 1995

We analyze number agreement in English in the version of Situation Semantics I put forth in [1]. We propose two theoretical notions, agreement target situations and relevance of situations. We propose that structural constraints shift the former in a way governed by the latter. Our analysis shows that number agreement reflects our knowledge about the structure of the world and the speaker's interests in the discourse.

• English Language & Literature Teaching
• /
• v.17 no.1
• /
• pp.147-163
• /
• 2011

In recent years, the relationship of language maintenance to culture and identity has received increased attention in the language acquisition and education fields. Korean immigrants in the U.S. form one of the biggest Asian groups and their language and cultural maintenance has been a major issue for both parents and ESL teachers. The present research is designed to investigate the cultural and social identities as well as the psychological investment factors that contribute to heritage language maintenance. Three Korean immigrant families in a small Midwest university town in the U.S. were surveyed and later interviewed. Issues and strategies concerning their children's Korean education in the U.S., coupled with the competing goal for the children to learn English were documented through parent interviews and interviews with school-aged focal children. Strategies and stances that facilitate or hinder both heritage and target language maintenance goals are presented along with participants' major reasons for heritage language maintenance in their homes and via Saturday schools. This work will assist ESL teachers and sociolinguists in situating both Korean student and parent goals in the context of shifting cultural and linguistic identities in countries where they have immigrated.

• Korean Journal of Cognitive Science
• /
• v.15 no.2
• /
• pp.43-53
• /
• 2004

The purpose of this study was to investigate the role of phonology in lexical access of bilingual processing for Korean-English bilinguals. Four experiments have been conducted in order to adjudicate the nonselective lexical access hypothesis, which argues simultaneous phonological activation of two bilingual languages, and the selective lexical access hypothesis. which argues phonological activation of only one bilingual language. The results showed that the Korean target word processing was significantly affected by the phonological manipulation of the English target word(Exp. 2). Similarly, the English target word processing showed the tendencies that it is affected by the phonological manipulation of the Korean prime word(Exp. 2). This results indicates that the phonological information of another bilingual language is automatically activated when we process one of bilingual languages, and the process of English which is the second language for most Korean, is phonologically activated.

• English Language & Literature Teaching
• /
• v.11 no.2
• /
• pp.91-110
• /
• 2005

Many Korean EFL (English as a Foreign Language) students do not have sufficient opportunity to develop cultural knowledge and information in their classrooms. EFL teachers also tend to ignore the teaching of culture. Even though culture is taught, it simply tends to deliver "fact-only" information from the viewpoint of a "tourist level rather than cultural awareness by comparing native with target cultural references. Teaching target cultural knowledge and information should be delivered within the native cultural frame, and teaching of culture must be an integral part of teaching and learning English. The research methodology was quantitative. Quantitative data was gathered from 83 Korean EFL teachers and 286 EFL students by questionnaire. Findings indicated that three of these independent variables (cultural inequality, English-only instruction, and Unoism) were significantly and inversely related to integration of culture.

• ETRI Journal
• /
• v.43 no.5
• /
• pp.775-786
• /
• 2021

To translate in real time, a simultaneous translation system should determine when to stop reading source tokens and generate target tokens corresponding to a partial source sentence read up to that point. However, conventional attention-based neural machine translation (NMT) models cannot produce translations with adequate latency in online scenarios because they wait until a source sentence is completed to compute alignment between the source and target tokens. To address this issue, we propose a reinforced learning (RL)-based attention mechanism, the reinforced attention mechanism, which allows a neural translation model to jointly train the stopping criterion and a partial translation model. The proposed attention mechanism comprises two modules, one to ensure translation quality and the other to address latency. Different from previous RL-based simultaneous translation systems, which learn the stopping criterion from a fixed NMT model, the modules can be trained jointly with a novel reward function. In our experiments, the proposed model has better translation quality and comparable latency compared to previous models.