• 재활복지
• /
• v.17 no.4
• /
• pp.371-382
• /
• 2013

Recently, the number of those who were transplanted cochlear implants increased as health insurance increases has expanded. Last six years between 2005 to 2009, patients who received a cochlear implant surgery were about 3,300 and number of cochlear implants in adults of them have shown growing aspects. In the case of young children, they actively participated auditory training program after cochlear implant surgery and the studies related to auditory training in child are many, but the studies related to auditory training in adults is insufficient. In this study, we perform the auditory training for the female adult (age 54) received cochlear implant after language acquisition used Ling 6 sounds test, standardized consonants, vowels and sentences listening test and word recognition and confirmation test. As a result after auditory training for 10 weeks, she identified all phonemes in Ling 6 sound test and performed close to 100% in standardized consonants, vowels and sentences listening tests. Also, she improved the ability of real-world environmental sound and real-world words identifications by 57-95%. The results of this study showed the need of auditory training program with systematic and effective planning and considering the characteristics of the individual for adults.

• Journal of Digital Convergence
• /
• v.17 no.11
• /
• pp.233-240
• /
• 2019

In order to predict and respond to cyber attacks, a number of security companies quickly identify the methods, types and characteristics of attack techniques and are publishing Security Intelligence Reports(SIRs) on them. However, the SIRs distributed by each company are huge and unstructured. In this paper, we propose a framework that uses five analytic techniques to formulate a report and extract key information in order to reduce the time required to extract information on large unstructured SIRs efficiently. Since the SIRs data do not have the correct answer label, we propose four analysis techniques, Keyword Extraction, Topic Modeling, Summarization, and Document Similarity, through Unsupervised Learning. Finally, has built the data to extract threat information from SIRs, analysis applies to the Named Entity Recognition (NER) technology to recognize the words belonging to the IP, Domain/URL, Hash, Malware and determine if the word belongs to which type We propose a framework that applies a total of five analysis techniques, including technology.

• Journal of Multimedia Information System
• /
• v.9 no.2
• /
• pp.113-120
• /
• 2022

Mongolian folk songs are inspired by Mongolian labor songs and are classified into long and short songs. Mongolian long songs have ancient origins, are rich in legends, and are a great source of folklore. So it was inscribed by UNESCO in 2008. Mongolian written literature is formed under the direct influence of oral literature. Mongolian long song has 3 classes: ayzam, suman, and besreg by their lyrics and structure. In ayzam long song, the world perfectly embodies the philosophical nature of world phenomena and the nature of human life. Suman long song has a wide range of topics such as the common way of life, respect for ancestors, respect for fathers, respect for mountains and water, livestock and animal husbandry, as well as the history of Mongolia. Besreg long songs are dominated by commanded and trained characters. In this paper, we proposed a method to classify their 3 types of long songs using machine learning, based on their lyrics structures without semantic information. We collected lyrics of over 80 long songs and extracted 11 features from every single song. The features are the name of a song, number of the verse, number of lines, number of words, general value, double value, elapsed time of verse, elapsed time of 5 words, and the longest elapsed time of 1 word, full text, and type label. In experimental results, our proposed features show on average 78% recognition rates in function type machine learning methods, to classify the ayzam, suman, and besreg classes.

• The Journal of Korean Medicine Ophthalmology and Otolaryngology and Dermatology
• /
• v.35 no.4
• /
• pp.172-180
• /
• 2022

Objectives : This study reports a 61-year-old man with sudden hearing loss and tinnitus whose symptoms improved remarkably after treatment with Korean medicine. Methods : The patient was treated with herbal medicine(gamisoyo-san, samul-tang gagambang, gongjindan, and yukmijihwang-hwan), acupuncture and moxibustion. Puretone audiometry(PTA), speech audiometry(SA), Korean tinnitus handicap inventory(K-THI), patient self-report, numeric rating scale(NRS) 11, and review of system(ROS) were performed. Results : After a month of treatment, PTA was improved from profound to mild and audiometric findings were improved in speech reception threshold and word recognition score. After 2 months of treatment, K-THI decreased from 92 to 28 and NRS 11 score decreased more than 6 in every item. Conclusions : These results demonstrate that Korean medicine could be effective sudden hearing loss and tinnitus.

• Journal of Digital Convergence
• /
• v.20 no.3
• /
• pp.75-83
• /
• 2022

Korea, which has the highest pedestrian fatality rate among OECD countries, is making efforts to improve the safe walking environment by enacting laws focusing on pedestrian. Spatial clustering was conducted with scan statistics after examining the social network data related to traffic accidents for children and seniors. The word cloud was used to examine people's recognition Campaigns for children and literature survey for seniors were in main concern. Naedang and Yongsan are the regions with the highest relative risk of weak pedestrian for children and seniors. On the contrary, Bongmu and Beomeo are the lowest relative risk region. Naedang-dong and Yongsan-dong of Daegu Metropolitan City were identified as vulnerable areas for pedestrian safety due to the high risk of pedestrian accidents for children and the elderly. This means that the scan statistics are effective in searching for traffic accident risk areas.

• Journal of Information Processing Systems
• /
• v.19 no.3
• /
• pp.289-301
• /
• 2023

Due to the development and dissemination of modern technology, anyone can easily communicate using services such as social network service (SNS) through a personal computer (PC) or smartphone. The development of these technologies has caused many beneficial effects. At the same time, bad effects also occurred, one of which was the spam problem. Spam refers to unwanted or rejected information received by unspecified users. The continuous exposure of such information to service users creates inconvenience in the user's use of the service, and if filtering is not performed correctly, the quality of service deteriorates. Recently, spammers are creating more malicious spam by distorting the image of spam text so that optical character recognition (OCR)-based spam filters cannot easily detect it. Fortunately, the level of transformation of image spam circulated on social media is not serious yet. However, in the mail system, spammers (the person who sends spam) showed various modifications to the spam image for neutralizing OCR, and therefore, the same situation can happen with spam images on social media. Spammers have been shown to interfere with OCR reading through geometric transformations such as image distortion, noise addition, and blurring. Various techniques have been studied to filter image spam, but at the same time, methods of interfering with image spam identification using obfuscated images are also continuously developing. In this paper, we propose a deep learning-based spam image detection model to improve the existing OCR-based spam image detection performance and compensate for vulnerabilities. The proposed model extracts text features and image features from the image using four sub-models. First, the OCR-based text model extracts the text-related features, whether the image contains spam words, and the word embedding vector from the input image. Then, the convolution neural network-based image model extracts image obfuscation and image feature vectors from the input image. The extracted feature is determined whether it is a spam image by the final spam image classifier. As a result of evaluating the F1-score of the proposed model, the performance was about 14 points higher than the OCR-based spam image detection performance.

• Annual Conference on Human and Language Technology
• /
• 2020.10a
• /
• pp.304-309
• /
• 2020

본 연구에서는 한국어 전자사전 DECO(Dictionnaire Electronique du COreen)와 다단어(Multi-Word Expressions: MWE) 개체명을 부분 패턴으로 기술하는 부분문법그래프(Local-Grammar Graph: LGG) 프레임에 기반하여 반자동으로 개체명주석 코퍼스 DecoNAC을 구축한 후, 이를 개체명 분석에 활용하고 또한 기계학습에 필요한 도메인별 학습 데이터로 활용하는 DecoNERO 개체명인식 플랫폼을 소개하는 데에 목적을 두었다. 최근 들어 좋은 성과를 보이는 것으로 보고되고 있는 기계학습 방법론들은 다양한 도메인을 기반으로한 대규모의 학습데이터를 필요로 한다. 본 연구에서는 정교하게 설계된 개체명 사전과 다단어 개체명 시퀀스에 대한 언어자원을 바탕으로 하는 반자동으로 학습데이터를 생성하는 방법론을 제안하였다. 본 연구에서 제안된 개체명주석 코퍼스 DecoNAC 기반 접근법의 성능을 실험하기 위해 온라인 뉴스 기사 텍스트를 바탕으로 실험을 진행하였다. 이 실험에서 DecoNAC을 적용한 경우, KoBERT 모델만으로 개체명을 인식한 결과에 비해 약 7.49%의 성능향상을 기대할 수 있음을 확인하였다.

• Journal of English Language & Literature
• /
• v.57 no.5
• /
• pp.787-812
• /
• 2011

This essay aims to explore Herman Melville's recognition and resolution of the vicious link between law and violence in his posthumous Billy Budd, Sailor (1924). In order to investigate the issues, this essay refers to Freud, Benjamin, Derrida, Lacan, and Žižek, all perceptive to the uncanny affinity of law and violence. Especially, Žižek's arguments of "superego" as an embodiment of cruel and destructive violence supplementing the official law and of "love" as an ethical possibility beyond the limit of the problematic law are introduced in this study to make Melville's reflection of the inseparableness between law and violence much clearer. John Claggart and Captain Vere embody the legal (superegoic) violence. Claggart even procurs secret enjoyment, in the name of maintaining positive law. Billy Budd discloses another violence defending his justness according to natural law. However, Melville suggests the possibility of suspending the problematic tie of law/violence through "love," as portrayed at the last part of the story. The two final words from Billy and Vere, as a sort of delayed dialogue between them after the event of their secret interview before Billy's hanging, suggest that they finally distance from the obscene nightly law of superego-respectively from outward punitiveness toward Vere and from inward punishment for Vere's excessive enforcement of Billy's hanging-and identify each other's lack as their own. Their love implicated in the last words is for the real other-in Lacan's sense-who discloses the constitutive lack or incompleteness of beings and aporia of the law. This essay's examination of Melville's representations about the superegoic violence as the (im-)possible condition of law and the possibility of withdrawing from it would help us recognize Billy Budd, Sailor as the author's own last word for the possible vision of love cutting the vicious knot of law/violence.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.659-663
• /
• 2023

이 논문은 한국어 음성인식 성능을 개선하고자 기존 음성인식 과정을 자모단위 음성인식 모델과 신경망 기반 자모 병합 모델 총 두 단계로 구성하였다. 한국어는 조합어 특성상 음성 인식에 필요한 음절 단위가 약 2900자에 이른다. 이는 학습 데이터셋에 자주 등장하지 않는 음절에 대해서 음성인식 성능을 저하시키고, 학습 비용을 높이는 단점이 있다. 이를 개선하고자 음절 단위의 인식이 아닌 51가지 자모 단위(ㄱ-ㅎ, ㅏ-ㅞ)의 음성인식을 수행한 후 자모 단위 인식 결과를 음절단위의 한글로 병합하는 과정을 수행할 수 있다[1]. 자모단위 인식결과는 초성, 중성, 종성을 고려하면 규칙 기반의 병합이 가능하다. 하지만 음성인식 결과에 잘못인식된 자모가 포함되어 있다면 최종 병합 결과에 오류를 생성하고 만다. 이를 해결하고자 신경망 기반의 자모 병합 모델을 제시한다. 자모 병합 모델은 분리되어 있는 자모단위의 입력을 완성된 한글 문장으로 변환하는 작업을 수행하고, 이 과정에서 음성인식 결과로 잘못인식된 자모에 대해서도 올바른 한글 문장으로 변환하는 오류 수정이 가능하다. 본 연구는 한국어 음성인식 말뭉치 KsponSpeech를 활용하여 실험을 진행하였고, 음성인식 모델로 Wav2Vec2.0 모델을 활용하였다. 기존 규칙 기반의 자모 병합 방법에 비해 제시하는 자모 병합 모델이 상대적 음절단위오류율(Character Error Rate, CER) 17.2% 와 단어단위오류율(Word Error Rate, WER) 13.1% 향상을 확인할 수 있었다.

• Korean Journal of Cognitive Science
• /
• v.35 no.2
• /
• pp.97-112
• /
• 2024

This study investigated how syllable-level variables such as syllable frequency, syllable (i.e. vowel) type, presence of final consonants (i.e. batchim) and syllable position influence naming of both words and pseudo-words. The results of the linear mixed-effects model analysis showed that, for words, naming time decreased as the frequency of the first syllable increased, and when the first syllable had a final consonant. Additionally, words were named more accurately when they had vertical vowels compared to horizontal vowels. For pseudo-words, naming time decreased and accuracy rate increased as the frequency of the first or the second syllable increased. Furthermore, pseudo-words were named more accurately when they had vertical vowels compared to horizontal vowels. These results suggest that while the frequency of the second syllable had differential effects between words and pseudo-words, the frequency of the first syllable and the syllable type had consistent effects for both words and pseudo-words. The implications of this study were discussed concerning visual word recognition processing.