• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.596-599
• /
• 2022

심층 학습은 지속적으로 발전하고 있으며, 최근에는 실제 사용자에게 제공되는 애플리케이션까지 확장되고 있다. 특히 자연어처리 분야에서는 대용량 언어 말뭉치를 기반으로 한 언어모델이 등장하면서 사람보다 높은 성능을 보이는 시스템이 개발되었다. 그러나 언어모델은 높은 컴퓨팅 파워를 요구하기 때문에 독립적인 소형 디바이스에서 제공할 수 있는 서비스에 적용하기 힘들다. 예를 들어 스캐너에서 제공할 수 있는 파일명 자동 부착 서비스는 하드웨어의 컴퓨팅 파워가 제한적이기 때문에 언어모델을 적용하기 힘들다. 또한, 활용할 수 있는 공개 데이터가 많지 않기 때문에, 데이터 구축에도 높은 비용이 요구된다. 따라서 본 논문에서는 컴퓨팅 파워에 비교적 독립적이고 학습 데이터가 필요하지 않은 비지도 학습을 활용하여 파일명 자동 부착 서비스를 위한 파일명 추출 방법을 제안한다. 실험은 681건의 문서 OCR 결과에 정답을 부착하여 수행했으며, ROUGE-L 기준 0.3352의 성능을 보였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• fall
• /
• pp.299-300
• /
• 2021

비대면 시대로 인한 화상 회의의 중요성이 높아졌다. 하지만 현재까지도 회의기록의 문서화 작업은 수작업으로 이루어지고 있어 시간과 인적자원이 많이 소모되고 있다. 본 논문은 기존 수작업으로 진행되는 회의 문서화 과정의 문제점을 개선하고자 한국어 키워드 추출을 활용한 인공지능 회의록 자동 생성 시스템을 제안한다. 회의 음성 파일을 기반으로 STT 기술을 활용한 회의 전문을 자동 생성하고 전문에 KR-WordRank 알고리즘을 적용해 키워드를 추출, Summary API를 사용하여 요약본을 생성한다. 최종 결과로 회의 전문과 키워드, 요약본이 담긴 PDF 형식의 회의록을 사용자에게 제공하여, 수기 회의록 작성 시 들이는 시간적, 인적 비용 절감을 돕는다.

• ETRI Journal
• /
• v.46 no.1
• /
• pp.35-47
• /
• 2024

This study introduces CR-M-SpanBERT, a coreference resolution (CR) model that utilizes multiple embedding-based span bidirectional encoder representations from transformers, for antecedent recognition in natural language (NL) text. Information extraction studies aimed to extract knowledge from NL text autonomously and cost-effectively. However, the extracted information may not represent knowledge accurately owing to the presence of ambiguous entities. Therefore, we propose a CR model that identifies mentions referring to the same entity in NL text. In the case of CR, it is necessary to understand both the syntax and semantics of the NL text simultaneously. Therefore, multiple embeddings are generated for CR, which can include syntactic and semantic information for each word. We evaluate the effectiveness of CR-M-SpanBERT by comparing it to a model that uses SpanBERT as the language model in CR studies. The results demonstrate that our proposed deep neural network model achieves high-recognition accuracy for extracting antecedents from NL text. Additionally, it requires fewer epochs to achieve an average F1 accuracy greater than 75% compared with the conventional SpanBERT approach.

• Journal of information and communication convergence engineering
• /
• v.22 no.1
• /
• pp.80-87
• /
• 2024

The Manchu language holds historical significance, but a complete dataset of Manchu script letters for training optical character recognition machine-learning models is currently unavailable. Therefore, this paper describes the process of creating a robust dataset of extracted Manchu script letters. Rather than performing automatic letter segmentation based on whitespace or the thickness of the central word stem, an image of the Manchu script was manually inspected, and one copy of the desired letter was selected as a region of interest. This selected region of interest was used as a template to match all other occurrences of the same letter within the Manchu script image. Although the dataset in this study contained only 4,000 images of five Manchu script letters, these letters were collected from twenty-eight writing styles. A full dataset of Manchu letters is expected to be obtained through this process. The collected dataset was normalized and trained using a simple convolutional neural network to verify its effectiveness.

• The Journal of the Korea Contents Association
• /
• v.11 no.12
• /
• pp.28-39
• /
• 2011

While the broadband multimedia technologies have been developing, the commercial market of digital contents has also been widely spreading. Most of all, digital cartoon market like internet cartoon has been rapidly large so video cartooning continuously has been researched because of lack and variety of cartoon. Until now, video cartooning system has been focused in non-photorealistic rendering and word balloon. But the meaningful frame extraction must take priority for cartooning system when applying in service. In this paper, we propose new automatic frame extraction method for video cartooning system. At frist, we separate video and audio from movie and extract features parameter like MFCC and ZCR from audio data. Audio signal is classified to speech, music and speech+music comparing with already trained audio data using GMM distributor. So we can set speech area. In the video case, we extract frame using general scene change detection method like histogram method and extract meaningful frames in the cartoon using face detection among the already extracted frames. After that, first of all existent face within speech area image transition frame extract automatically. Suitable frame about movie cartooning automatically extract that extraction image transition frame at continuable period of time domain.

• The Journal of the Acoustical Society of Korea
• /
• v.23 no.2
• /
• pp.180-186
• /
• 2004

The presence of noise in speech signals degrades the performance of recognition systems in which there are mismatches between the training and test environments. To make a speech recognizer robust, it is necessary to compensate these mismatches. In this paper, we studied about an improvement of stochastic feature extraction based on band-SNR for robust speech recognition. At first, we proposed a modified version of the multi-band spectral subtraction (MSS) method which adjusts the subtraction level of noise spectrum according to band-SNR. In the proposed method referred as M-MSS, a noise normalization factor was newly introduced to finely control the over-estimation factor depending on the band-SNR. Also, we modified the architecture of the stochastic feature extraction (SFE) method. We could get a better performance when the spectral subtraction was applied in the power spectrum domain than in the mel-scale domain. This method is denoted as M-SFE. Last, we applied the M-MSS method to the modified stochastic feature extraction structure, which is denoted as the MMSS-MSFE method. The proposed methods were evaluated on isolated word recognition under various noise environments. The average error rates of the M-MSS, M-SFE, and MMSS-MSFE methods over the ordinary spectral subtraction (SS) method were reduced by 18.6%, 15.1%, and 33.9%, respectively. From these results, we can conclude that the proposed methods provide good candidates for robust feature extraction in the noisy speech recognition.

• Journal of KIISE:Software and Applications
• /
• v.30 no.1_2
• /
• pp.173-183
• /
• 2003

Morphological analysis is the most widely used method for extracting nouns from Korean texts. For every Eojeol, in order to extract nouns from it, a morphological analyzer performs frequent dictionary lookup and applies many morphonological rules, therefore it requires many operations. Moreover, a morphological analyzer generates all the possible morphological interpretations (sequences of morphemes) of a given Eojeol, which may by unnecessary from the noun extraction`s point of view. To reduce unnecessary computation of morphological analysis from the noun extraction`s point of view, this paper proposes a method for Korean noun extraction considering noun occurrence characteristics. Noun patterns denote conditions on which nouns are included in an Eojeol or not, which are positive cues or negative cues, respectively. When using the exclusive information as the negative cues, it is possible to reduce the search space of morphological analysis by ignoring Eojeols not including nouns. Post-noun syllable sequences(PNSS) as the positive cues can simply extract nouns by checking the part of the Eojeol preceding the PNSS and can guess unknown nouns. In addition, morphonological information is used instead of many morphonological rules in order to recover the lexical form from its altered surface form. Experimental results show that the proposed method can speed up without losing accuracy compared with other systems based on morphological analysis.

• The KIPS Transactions:PartB
• /
• v.14B no.6
• /
• pp.453-460
• /
• 2007

In this paper, we present a method to extract the text lines in poorly structured documents. The text lines may have different orientations, considerably curved shapes, and there are possibly a few wide inter-word gaps in a text line. Those text lines can be found in posters, blocks of addresses, artistic documents. Our method based on the traditional perceptual grouping but we develop novel solutions to overcome the problems of insufficient seed points and vaned orientations un a single line. In this paper, we assume that text lines contained tone connected components, in which each connected components is a set of black pixels within a letter, or some touched letters. In our scheme, the connected components closer than an iteratively incremented threshold will make together a chain. Elongate chains are identified as the seed chains of lines. Then the seed chains are extended to the left and the right regarding the local orientations. The local orientations will be reevaluated at each side of the chains when it is extended. By this process, all text lines are finally constructed. The proposed method is good for extraction of the considerably curved text lines from logos and slogans in our experiment; 98% and 94% for the straight-line extraction and the curved-line extraction, respectively.

• The Journal of the Acoustical Society of Korea
• /
• v.20 no.4
• /
• pp.35-41
• /
• 2001

This paper describes Linear Discriminant Analysis and common vector extraction for speech recognition. Voice signal contains psychological and physiological properties of the speaker as well as dialect differences, acoustical environment effects, and phase differences. For these reasons, the same word spelled out by different speakers can be very different heard. This property of speech signal make it very difficult to extract common properties in the same speech class (word or phoneme). Linear algebra method like BT (Karhunen-Loeve Transformation) is generally used for common properties extraction In the speech signals, but common vector extraction which is suggested by M. Bilginer et at. is used in this paper. The method of M. Bilginer et al. extracts the optimized common vector from the speech signals used for training. And it has 100％ recognition accuracy in the trained data which is used for common vector extraction. In spite of these characteristics, the method has some drawback-we cannot use numbers of speech signal for training and the discriminant information among common vectors is not defined. This paper suggests advanced method which can reduce error rate by maximizing the discriminant information among common vectors. And novel method to normalize the size of common vector also added. The result shows improved performance of algorithm and better recognition accuracy of 2％ than conventional method.

• Journal of Intelligence and Information Systems
• /
• v.28 no.1
• /
• pp.155-174
• /
• 2022

From January 2020 to October 2021, more than 500,000 academic studies related to COVID-19 (Coronavirus-2, a fatal respiratory syndrome) have been published. The rapid increase in the number of papers related to COVID-19 is putting time and technical constraints on healthcare professionals and policy makers to quickly find important research. Therefore, in this study, we propose a method of extracting useful information from text data of extensive literature using LDA and Word2vec algorithm. Papers related to keywords to be searched were extracted from papers related to COVID-19, and detailed topics were identified. The data used the CORD-19 data set on Kaggle, a free academic resource prepared by major research groups and the White House to respond to the COVID-19 pandemic, updated weekly. The research methods are divided into two main categories. First, 41,062 articles were collected through data filtering and pre-processing of the abstracts of 47,110 academic papers including full text. For this purpose, the number of publications related to COVID-19 by year was analyzed through exploratory data analysis using a Python program, and the top 10 journals under active research were identified. LDA and Word2vec algorithm were used to derive research topics related to COVID-19, and after analyzing related words, similarity was measured. Second, papers containing 'vaccine' and 'treatment' were extracted from among the topics derived from all papers, and a total of 4,555 papers related to 'vaccine' and 5,971 papers related to 'treatment' were extracted. did For each collected paper, detailed topics were analyzed using LDA and Word2vec algorithms, and a clustering method through PCA dimension reduction was applied to visualize groups of papers with similar themes using the t-SNE algorithm. A noteworthy point from the results of this study is that the topics that were not derived from the topics derived for all papers being researched in relation to COVID-19 (