• Proceedings of the KSPS conference
• /
• 2006.05a
• /
• pp.37-40
• /
• 2006

In this paper, we propose a VQ codebook design of speech recognition feature parameters in order to improve the performance of a distributed speech recognition system. For the context-dependent HMMs, a VQ codebook should be correlated with phonetic distributions in the training data for HMMs. Thus, we focus on a selection method of training data based on phonetic distribution instead of using all the training data for an efficient VQ codebook design. From the speech recognition experiments using the Aurora 4 database, the distributed speech recognition system employing a VQ codebook designed by the proposed method reduced the word error rate (WER) by 10% when compared with that using a VQ codebook trained with the whole training data.

• Journal of The Korean Association For Science Education
• /
• v.14 no.2
• /
• pp.143-158
• /
• 1994

This study aims to identify the characteristics of gas phase problem solving of college freshmen. Four students were participated in this study and solved the problem by using think-aloud method. The thinking processes were recorded and transferred into protocols. Problem solving stage, the ratio spended in each solving stage, solving strategy, misconceptions, and errors were identified and discussed. The relationships between students' belief system about chemistry problem solving and problem solving characteristics were also investigated. The results were as follows: 1. Students felt that chemical equation problem was easier than word problem or pictorial problem. 2. When students had declarative knowledge and procedural knowledge required by given problem, their confidence level and formula selection were not changed by redundunt information in the problem. 3. When the problem seemed to be difficult, students tended to use the Means-End or Random strategy. 4. In complicated problems, students spent longer time for problem apprehension and planning. In familiar problems, students spent rather short time for planning. 5. Students spent more time for overall problem solving process in case of using Means-End or Random strategy than using Knowledge-Development strategy.

• The KIPS Transactions:PartB
• /
• v.12B no.3 s.99
• /
• pp.365-368
• /
• 2005

We propose a speech act analysis method for Korean dialogue using Support Vector Machines (SVM). We use a lexical form of a word, its part of speech (POS) tags, and bigrams of POS tags as sentence features and the contexts of the previous utterance as context features. We select informative features by Chi square statistics. After training SVM with the selected features, SVM classifiers determine the speech act of each utterance. In experiment, we acquired overall $90.54\%$ of accuracy with dialogue corpus for hotel reservation domain.

• Proceedings of the KOSOMBE Conference
• /
• v.1997 no.11
• /
• pp.67-70
• /
• 1997

In this paper, we evaluated the character of speech of hearing impaired person (HIP) who acquire his hearing loss after the youth. It is usually observed that severe HIP decreased not only speech perception but also vocalization. so there is a need for sensitive and quantitative measures or the assesment of the speech of the HIP to serve both diagnostic and prognosic purposes, 7 HIP and 12 normal hearing person(NHP) were studied with pure tone test and speaking test using word/sentence table which consists of vowel(a:), mono and two syllables and a sentence. we analyzed formant frequency, pitch, sound intensity, speech duration of HIP and NHP speech. According to the results, in the HIP's speech we find that formant frequency was shifted, first-formant prominence was reduced, the dynamic range of sound intensity was decreased, speech duration was prolonged. In the next, we expect the correlation between hearing and speech character of HIP is cleared through analysis of more acoustic parameters and precise selection of HIP group.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.16 no.5 s.110
• /
• pp.457-469
• /
• 2006

In automotive engineering, the brand sound is one of the important advantage strategies in a car company. For the design of brand sound, the selection of descriptive word for a car sound is one of major works in automotive sound quality research. In this paper, booming and rumbling sound, which are professional words used by sound and vibration engineers are used for the design of brand sound. We employed sound quality metrics, which are used in the psychoacoustics. By most research results, the relationship between subjective evaluations and sound quality metrics has nonlinear characteristics. In order to correlate these subjective evaluations with sound quality metrics, the artificial neural network technology has been applied to two-dimensional sound quality index for a passenger car. These indexes are used for 46 passenger cars, which are samples of the famous cars around the world. Also a preference evaluation for car sound was carried out by sound and vibration engineers. We coupled this preference with booming and rumbling sounds by using artificial neural network. In future, the two dimensional sound and preference index will be very useful to develop brand sound in passenger cars.

• Annual Conference on Human and Language Technology
• /
• 1999.10e
• /
• pp.209-214
• /
• 1999

패턴기반기계번역방식은 원시언어패턴과 그에 대한 대역언어패턴들의 쌍을 이용하여 구문분석과 변환을 수행하는 기계번역방식이다. 패턴기반 기계번역방식은 번역할 때 발생하는 애매성을 해소하기 위해 패턴의 길이를 문장단위까지 늘이기 때문에, 패턴의 수가 급증하는 문제점을 가진다. 본 논문에서는 패턴의 단위를 구단위로 한정시킬 때 발생하는 애매성을 해소하는 방법으로 시소러스를 기반으로 한 두단계 대역어 선택 방식을 제안함으로써 효과적으로 애매성을 감소시키면서 패턴의 길이를 줄이는 모델을 제시한다. 두단계 대역어 선택 방식은 원시언어의 한 패턴에 대해 여러 가능한 목적언어의 대역패턴들이 있을 때, 첫 번째 단계에서는 원시언어 내에서의 제약조건에 맞는 몇가지 대역패턴들을 선택하고, 두번째 단계에서는 목적언어 내에서의 제약조건에 가장 적합한 하나의 대역패턴을 선택하는 방식이다. 또한 본 논문에서는 이와 같은 모델에서 패턴의 수가 코퍼스의 증가에 따른 수렴가능성을 논한다.

• Journal of Engineering Education Research
• /
• v.12 no.3
• /
• pp.96-106
• /
• 2009

The purpose of this study was to suggest possibility of adaptation of course-embedded assessment for program outcome in engineering education through literature review on concept, benefits, adaptation process, case study in higher education of course-embedded assessment. The model of course-embedded assessment was given with consideration for each step, that is, set up of program outcomes, check of related courses, selection of assessment tools, set up of performance criteria, implementation of course-embedded assessment, and analysis of results. Some prerequisites were given to set up course-embedded assessment successfully in engineering education field of Korea.

• The Journal of Industrial Distribution & Business
• /
• v.4 no.2
• /
• pp.41-49
• /
• 2013

Purpose - Currently, the distribution environment has been changing rapidly because of the advent of super-supermarkets (SSM),which adopt the "price break" and "lowest price" strategies. Specifically, a diversification of the distribution environment allowed customers to reasonably decide where to shop. With a broadened range of criteria governing the choice of a store, determining changes in the distribution environment should be based on customer satisfaction surrounding the service levels of SSMs. This study examines the effects of service quality and product quality of SSMs on customer satisfaction levels, and analyzes whether these factors affect customers' revisit intention directly. Research design and methodology - A survey method was used and 127 SSM customers responded to the questionnaires that were developed to measure service quality, product quality, customer satisfaction, and store revisit intention. A multi-regression analysis was used to test the hypotheses. Results - Except hypothesis 2, the other seven hypotheses were supported. Hypothesis 2 suggested a positive relationship between product factors and customer satisfaction. Conclusions - This study found a systematic effect of service factors and product factors on customer satisfaction and word-of-mouth, and consequently on store revisit intention.

• The KIPS Transactions:PartB
• /
• v.17B no.3
• /
• pp.249-254
• /
• 2010

We propose an automatic spam filter for e-mail data using Support Vector Machines(SVM). We use a lexical form of a word and its part of speech(POS) tags as features and select features by chi square statistics. We represent each feature by TF(text frequency), TF-IDF, and binary weight for experiments. After training SVM with the selected features, SVM classifies each e-mail as spam or not. In experiment, the selected features improve the performance of our system and we acquired overall 98.9% of accuracy with TREC05-p1 spam corpus.

• Journal of Korean society of Dental Hygiene
• /
• v.21 no.5
• /
• pp.639-644
• /
• 2021

Objectives: The purpose of this study was to analyze the factors that affected the number of citations for articles published in the Journal of Korean Society of Dental Hygiene based on previous studies. Methods: Information on papers including the number of citations was collected using a web crawling technique. The effect of the number of author keywords, the number of Medical Subject Headings (MeSH) keywords, MeSH match rate, abstract word count and keyword-abstract ratio on the number of citations was analyzed by multiple regression analysis. Results: The use of the MeSH keyword did not have a significant effect on the number of citations. Among the other factors, only the keyword-abstract ratio was statistically significant. Conclusions: Select a topic of constant interest in the field, write the title in detail using colons or asterisks if necessary, and do not repeat the words used in the title in keywords. Select specific keywords deeply related to the topic. In particular, choice words or phrases that are frequently used in the abstract. If the MeSH keyword selection contradicts the previous strategies, boldly give up the MeSH keyword.