• Title/Summary/Keyword: BLEU

Search Result 72, Processing Time 0.024 seconds

Molecular Analysis of Alternative Transcripts of the Equine Cordon-Bleu WH2 Repeat Protein-Like 1 (COBLL1) Gene

  • Park, Jeong-Woong;Jang, Hyun-Jun;Shin, Sangsu;Cho, Hyun-Woo;Choi, Jae-Young;Kim, Nam-Young;Lee, Hak-Kyo;Do, Kyong-Tak;Song, Ki-Duk;Cho, Byung-Wook
    • Asian-Australasian Journal of Animal Sciences
    • /
    • v.28 no.6
    • /
    • pp.870-875
    • /
    • 2015
  • The purpose of this study was to investigate the alternative splicing in equine cordon-bleu WH2 repeat protein-like 1 (COBLL1) gene that was identified in horse muscle and blood leukocytes, and to predict functional consequences of alternative splicing by bioinformatics analysis. In a previous study, RNA-seq analysis predicted the presence of alternative spliced isoforms of equine COBLL1, namely COBLL1a as a long form and COBLL1b as a short form. In this study, we validated two isoforms of COBLL1 transcripts in horse tissues by the real-time polymerase chain reaction, and cloned them for Sanger sequencing. The sequencing results showed that the alternative splicing occurs at exon 9. Prediction of protein structure of these isoforms revealed three putative phosphorylation sites at the amino acid sequences encoded in exon 9, which is deleted in COBLL1b. In expression analysis, it was found that COBLL1b was expressed ubiquitously and equivalently in all the analyzed tissues, whereas COBLL1a showed strong expression in kidney, spinal cord and lung, moderate expression in heart and skeletal muscle, and low expression in thyroid and colon. In muscle, both COBLL1a and COBLL1b expression decreased after exercise. It is assumed that the regulation of COBLL1 expression may be important for regulating glucose level or switching of energy source, possibly through an insulin signaling pathway, in muscle after exercise. Further study is warranted to reveal the functional importance of COBLL1 on athletic performance in race horses.

Sign Language Dataset Built from S. Korean Government Briefing on COVID-19 (대한민국 정부의 코로나 19 브리핑을 기반으로 구축된 수어 데이터셋 연구)

  • Sim, Hohyun;Sung, Horyeol;Lee, Seungjae;Cho, Hyeonjoong
    • KIPS Transactions on Software and Data Engineering
    • /
    • v.11 no.8
    • /
    • pp.325-330
    • /
    • 2022
  • This paper conducts the collection and experiment of datasets for deep learning research on sign language such as sign language recognition, sign language translation, and sign language segmentation for Korean sign language. There exist difficulties for deep learning research of sign language. First, it is difficult to recognize sign languages since they contain multiple modalities including hand movements, hand directions, and facial expressions. Second, it is the absence of training data to conduct deep learning research. Currently, KETI dataset is the only known dataset for Korean sign language for deep learning. Sign language datasets for deep learning research are classified into two categories: Isolated sign language and Continuous sign language. Although several foreign sign language datasets have been collected over time. they are also insufficient for deep learning research of sign language. Therefore, we attempted to collect a large-scale Korean sign language dataset and evaluate it using a baseline model named TSPNet which has the performance of SOTA in the field of sign language translation. The collected dataset consists of a total of 11,402 image and text. Our experimental result with the baseline model using the dataset shows BLEU-4 score 3.63, which would be used as a basic performance of a baseline model for Korean sign language dataset. We hope that our experience of collecting Korean sign language dataset helps facilitate further research directions on Korean sign language.

Evaluation Method of Machine Translation System (기계번역 성능평가를 위한 핵심어 전달율 측정방안)

  • Yu, Cho-Rong;Lee, Young-Jik;Park, Jun
    • Annual Conference on Human and Language Technology
    • /
    • 2003.10d
    • /
    • pp.241-245
    • /
    • 2003
  • 본 논문은 기계번역 시스템의 성능평가를 위한 '핵심어 전달율 측정' 방안에 대해서 기술한다. 기계번역 시스템의 성능평가는 두 가지 측면으로 고려될 수 있다. 첫 번째는 객관적인 평가로 IBM에서 주창한 BLEU score 측정이나 NIST의 NIST score 측정이 그 예이다. 객관적인 평가는 평가자의 주관적인 판단이나 언어적인 특성을 배제한 방법으로 프로그램을 통해 자동으로 fluency와 adequacy를 측정하여 성능을 평가한다. 다음은 주관적인 평가이다. 주관적인 평가는 평가자의 평가를 통해 번역의 품질을 평가하는 방법이다. 주관적 평가 방법의 대표적인 것으로는 NESPOLE이나 LDC가 있다. 주관적인 평가는 평가자의 정확한 판단으로 신뢰할만한 성능평가 결과를 도출하지만, 시간과 비용이 많이 들고, 재사용할 수 없다는 단점이 있다. 본 논문에서는 이러한 문제를 해결하기 위해, 번역대상 문장에서 핵심어를 추출하고, 그 핵심어가 기계번역 시스템의 수행결과에 전달된 정도를 자동으로 측정하는 새로운 평가방법인 '핵심어 전달율 측정' 방안을 제안한다. 이는 성능평가의 비용과 시간을 절약하고, 주관적 평가와 유사한 신뢰성 있는 평가결과를 얻을 수 있는 좋은 지표가 될 수 있을 것으로 기대한다.

  • PDF

Embedded clause extraction and restoration for the performance enhancement in Korean-Vietnamese statistical machine translation (한베 통계기계번역의 성능 향상을 위한 내포문 추출 및 복원 기법)

  • Cho, Seung-Woo;Kim, Young-Gil;Kwon, Hong-Seok;Lee, Eui-Hyun;Lee, Won-Ki;Cho, Hyung-Mi;Lee, Jong-Hyeok
    • 한국어정보학회:학술대회논문집
    • /
    • 2016.10a
    • /
    • pp.280-284
    • /
    • 2016
  • 본 논문에서는 기호로 둘러싸인 내포문이 포함된 문장의 번역 성능을 높이는 방법을 제안한다. 입력 문장에서 내포문을 추출하여 여러 문장으로 나타내고, 각각의 문장들을 번역한다. 그리고 번역된 문장들을 복원정보를 활용하여 최종 번역 문장을 생성한다. 이러한 방법론은 입력 문장의 길이를 줄여주며, 그로 인하여 문장 구조가 단순해져 번역 품질이 향상된다. 본 논문에서는 한국어-베트남어 통계 기반 번역기에 대하여 제안한 방법론을 적용하고 실험하였다. 그 결과 BLEU 점수가 약 1.5 향상된 것을 확인할 수 있었다.

  • PDF

Generative Multi-Turn Chatbot Using Generative Adversarial Network (생성적 적대적 신경망을 이용한 생성기반 멀티턴 챗봇)

  • Kim, Jintae;Kim, Harksoo;Kwon, Oh-Woog;Kim, Young-Gil
    • Annual Conference on Human and Language Technology
    • /
    • 2018.10a
    • /
    • pp.25-30
    • /
    • 2018
  • 기존의 검색 기반 챗봇 시스템과 다르게 생성 기반 챗봇 시스템은 사전에 정의된 응답에 의존하지 않고 채팅 말뭉치를 학습한 신경망 모델을 사용하여 응답을 생성한다. 생성 기반 챗봇 시스템이 사람과 같이 자연스러운 응답을 생성하려면 이전 문맥을 반영해야 할 필요가 있다. 기존 연구에서는 문맥을 반영하기 위해 이전 문맥과 입력 발화를 통합하여 하나의 벡터로 표현했다. 이러한 경우 이전 문맥과 입력 발화가 분리되어 있지 않아 이전 문맥이 필요하지 않는 경우 잡음으로 작용할 수 있다. 본 논문은 이러한 문제를 해결하기 위해 입력 발화와 이전 문맥을 각각의 벡터로 표현하는 방법을 제안한다. 또한 생성적 적대적 신경망을 통해 챗봇 시스템을 보강하는 방법을 제안한다. 채팅 말뭉치(55,000 개의 학습 데이터, 5,000개의 검증 데이터, 5,260 개의 평가 데이터)를 사용한 실험에서 제안한 문맥 반영 방법과 생성적 적대적 신경망을 통한 챗봇 시스템 보강 방법은 BLEU와 임베딩 기반 평가의 성능 향상에 도움을 주었다.

  • PDF

Study on Korean Fermented Sauce applied to Western Cuisine - Focused on Red Pepper Paste, Soybean Paste, Soy Sauce and Vinegar - (한국 발효 소스의 서양요리 적용에 대한 연구 - 고추장, 된장, 간장, 식초를 중심으로 -)

  • Kim, Jihyung;Yoo, Eunyi
    • Journal of the East Asian Society of Dietary Life
    • /
    • v.27 no.2
    • /
    • pp.223-234
    • /
    • 2017
  • The purpose of this study was to determine the possibilities of Korean fermented sauces including red pepper paste, soybean paste, soy sauce, and vinegar as ingredients for Western cuisine. Western cuisine professionals from US and Europe were interviewed for their experienced opinions. To classify the categories, the selected statements were given to other groups of foreign chefs, Korean cuisine professionals and students majoring culinary arts. The first category pointed out that Korean fermented sauces are healthy with 'umami' taste using only natural ingredients. They believe it has high possibilities of matching with many of other foods and also has unique tastes. Korean cuisine professionals were mostly occupied in this category. The second category had negative opinions matching with Western cuisines since Korean fermented sauces are rough and have a strong taste & smell. This category had many Western cuisine professionals. The last category was composed of mainly students majoring in culinary arts. They pointed out that Korean fermented sauces use natural ingredients and have a unique flavor with long-term shelf life. Use of Q methodology was significantly different from previous studies researched by quantitative methods especially for the Korea food service industry.

Customer Loyalty and Perception Differences in Relational Benefit: Focusing on Restaurant Industries (외식고객의 충성도 분류에 따른 관계편익 지각 차이에 대한 연구)

  • Kim, Hyungmin;Yoon, Jiyoung
    • Culinary science and hospitality research
    • /
    • v.24 no.1
    • /
    • pp.50-62
    • /
    • 2018
  • The purpose of this study was to overview the meaning of customer loyalty to segment customers based on their loyalty and to analyze the difference of loyal customers' perception of relational benefits in the restaurant industries. A self-administered questionnaire was distributed to 500 adults with dining experience at restaurants. Participants were given a brief description of loyalty and were made to choose a specific restaurant they felt loyal to and one with no loyalty. Attitudinal and behavioral loyalty were used in cluster analysis resulting 4 cluster groups. Each group was named true, spurious, latent, and low loyalty. After the groups were separated, ANOVA was used to see if the score of perceived relational benefit showed difference. All four relational benefit including social, psychological, economic, and customization benefit showed significant difference(p<.001). True loyal customers perceived relational benefit as the highest while low loyal customers showed the lowest. For latent and spurious loyal customers, it was found that latent loyal customers showed higher perception than spurious customers.

Understanding recurrent neural network for texts using English-Korean corpora

  • Lee, Hagyeong;Song, Jongwoo
    • Communications for Statistical Applications and Methods
    • /
    • v.27 no.3
    • /
    • pp.313-326
    • /
    • 2020
  • Deep Learning is the most important key to the development of Artificial Intelligence (AI). There are several distinguishable architectures of neural networks such as MLP, CNN, and RNN. Among them, we try to understand one of the main architectures called Recurrent Neural Network (RNN) that differs from other networks in handling sequential data, including time series and texts. As one of the main tasks recently in Natural Language Processing (NLP), we consider Neural Machine Translation (NMT) using RNNs. We also summarize fundamental structures of the recurrent networks, and some topics of representing natural words to reasonable numeric vectors. We organize topics to understand estimation procedures from representing input source sequences to predict target translated sequences. In addition, we apply multiple translation models with Gated Recurrent Unites (GRUs) in Keras on English-Korean sentences that contain about 26,000 pairwise sequences in total from two different corpora, colloquialism and news. We verified some crucial factors that influence the quality of training. We found that loss decreases with more recurrent dimensions and using bidirectional RNN in the encoder when dealing with short sequences. We also computed BLEU scores which are the main measures of the translation performance, and compared them with the score from Google Translate using the same test sentences. We sum up some difficulties when training a proper translation model as well as dealing with Korean language. The use of Keras in Python for overall tasks from processing raw texts to evaluating the translation model also allows us to include some useful functions and vocabulary libraries as well.

Embedded clause extraction and restoration for the performance enhancement in Korean-Vietnamese statistical machine translation (한베 통계기계번역의 성능 향상을 위한 내포문 추출 및 복원 기법)

  • Cho, Seung-Woo;Kim, Young-Gil;Kwon, Hong-Seok;Lee, Eui-Hyun;Lee, Won-Ki;Cho, Hyung-Mi;Lee, Jong-Hyeok
    • Annual Conference on Human and Language Technology
    • /
    • 2016.10a
    • /
    • pp.280-284
    • /
    • 2016
  • 본 논문에서는 기호로 둘러싸인 내포문이 포함된 문장의 번역 성능을 높이는 방법을 제안한다. 입력 문장에서 내포문을 추출하여 여러 문장으로 나타내고, 각각의 문장들을 번역한다. 그리고 번역된 문장들을 복원정보를 활용하여 최종 번역 문장을 생성한다. 이러한 방법론은 입력 문장의 길이를 줄여주며, 그로 인하여 문장 구조가 단순해져 번역 품질이 향상된다. 본 논문에서는 한국어-베트남어 통계 기반 번역기에 대하여 제안한 방법론을 적용하고 실험하였다. 그 결과 BLEU 점수가 약 1.5 향상된 것을 확인할 수 있었다.

  • PDF

A Study on the Parents' Perceptions of Children's Favorite Foods (어린이 기호식품에 대한 학부모 인식 조사)

  • Jung, Ji-Hye;Song, Kyung-Hee;Yoon, Ji-Young
    • Korean Journal of Community Nutrition
    • /
    • v.14 no.1
    • /
    • pp.67-76
    • /
    • 2009
  • The purpose of this study was to investigate the parents' perceptions of children's favorite foods. Mothers of elementary school students in Seoul were asked in a survey about their perceptions of children's favorite foods and their opinions of the related policy. Respondents pointed out the problems of children's favorite foods including insufficient sanitation, concerns with food additives, untrustworthy manufacturer, unsafe food distribution system and overuse of MSG. Overall hazardous perceptions of children's favorite foods were 2.71 out of 4.00. Most respondents believed that the children's favorite foods contained some harmful ingredients or over nutrients, and 69.2% of those respondents knew exactly which ingredients may cause children's health problems. The hazardous perception of chocolate, yogurt, sport drink and fruit drink were low compared to others, whereas hazardous perceptions and accuracy were high in candies, icebars, hamburgers and pizza, In terms of comprehensive countermeasures against unsafe children's foods, the respondents perceived that the establishment of standard amounts of nutrient value and food additives was the most important issue.