• The Journal of the Korea institute of electronic communication sciences
• /
• v.19 no.1
• /
• pp.113-118
• /
• 2024

In this paper, we propose an optimal wavelet in a system for canceling background noise of acoustic signals. This system performed Discrete Wavelet Transform(DWT) instead of the existing Short Time Fourier Transform(STFT) and then improved noise cancellation performance through a deep learning process. DWT functions as a multi-resolution band-pass filter and obtains transformation parameters by time-shifting the parent wavelet at each level and using several wavelets whose sizes are scaled. Here, the noise cancellation performance of several wavelets was tested to select the most suitable mother wavelet for analyzing the speech. In this study, to verify the performance of the noise cancellation system for various wavelets, a simulation program using Tensorflow and Keras libraries was created and simulation experiments were performed for the four most commonly used wavelets. As a result of the experiment, the case of using Haar or Daubechies wavelets showed the best noise cancellation performance, and the mean square error(MSE) was significantly improved compared to the case of using other wavelets.

• Journal of Practical Engineering Education
• /
• v.16 no.3_spc
• /
• pp.343-349
• /
• 2024

Recently, the intercultural approach, which aims to resolve social conflicts in multicultural societies through cultural encounters and dialogue, has been actively discussed. Intercultural education aims to foster smooth relationships and improve communication skills through interactions among various cultural groups. Analysis of previous studies has revealed the need for research on intercultural civic education programs targeted at adults. Therefore, this study was designed to develop an intercultural citizenship education program and to analyze its effects. For the study, previous research on interculturalism, intercultural citizenship education, and intercultural citizenship education programs was comprehensively analyzed. The developed intercultural citizenship education program was categorized into five major themes: understanding oneself and others, racism and hate speech, overcoming conflict and discrimination, breaking away from prejudice and stereotypes, and the future of our country. Subsequently, the program was implemented with 64 total second-year middle school students, 37 in the experimental group, and 27 in the control group, at an accredited lifelong educational institution in M City, South Jeolla Province. To analyze the effects of the education, an intercultural competency scale was used. The results showed that the intercultural citizenship education program for adult students was effective in conflict management, respect, communication, and reflection competencies, all at a statistically significant level. Finally, the authors compared the study results with previous research and discussed the findings. Findings indicated the necessity of enhancing multicultural citizenship awareness among adults and expanding teacher training in intercultural education. The program effectively improved intercultural competence among adult students, highlighting the importance of tailored educational content and active participation in discussions. Future research should ensure a balanced gender ratio among participants.

• Journal of Intelligence and Information Systems
• /
• v.23 no.2
• /
• pp.71-88
• /
• 2017

Language models were originally developed for speech recognition and language processing. Using a set of example sentences, a language model predicts the next word or character based on sequential input data. N-gram models have been widely used but this model cannot model the correlation between the input units efficiently since it is a probabilistic model which are based on the frequency of each unit in the training set. Recently, as the deep learning algorithm has been developed, a recurrent neural network (RNN) model and a long short-term memory (LSTM) model have been widely used for the neural language model (Ahn, 2016; Kim et al., 2016; Lee et al., 2016). These models can reflect dependency between the objects that are entered sequentially into the model (Gers and Schmidhuber, 2001; Mikolov et al., 2010; Sundermeyer et al., 2012). In order to learning the neural language model, texts need to be decomposed into words or morphemes. Since, however, a training set of sentences includes a huge number of words or morphemes in general, the size of dictionary is very large and so it increases model complexity. In addition, word-level or morpheme-level models are able to generate vocabularies only which are contained in the training set. Furthermore, with highly morphological languages such as Turkish, Hungarian, Russian, Finnish or Korean, morpheme analyzers have more chance to cause errors in decomposition process (Lankinen et al., 2016). Therefore, this paper proposes a phoneme-level language model for Korean language based on LSTM models. A phoneme such as a vowel or a consonant is the smallest unit that comprises Korean texts. We construct the language model using three or four LSTM layers. Each model was trained using Stochastic Gradient Algorithm and more advanced optimization algorithms such as Adagrad, RMSprop, Adadelta, Adam, Adamax, and Nadam. Simulation study was done with Old Testament texts using a deep learning package Keras based the Theano. After pre-processing the texts, the dataset included 74 of unique characters including vowels, consonants, and punctuation marks. Then we constructed an input vector with 20 consecutive characters and an output with a following 21st character. Finally, total 1,023,411 sets of input-output vectors were included in the dataset and we divided them into training, validation, testsets with proportion 70:15:15. All the simulation were conducted on a system equipped with an Intel Xeon CPU (16 cores) and a NVIDIA GeForce GTX 1080 GPU. We compared the loss function evaluated for the validation set, the perplexity evaluated for the test set, and the time to be taken for training each model. As a result, all the optimization algorithms but the stochastic gradient algorithm showed similar validation loss and perplexity, which are clearly superior to those of the stochastic gradient algorithm. The stochastic gradient algorithm took the longest time to be trained for both 3- and 4-LSTM models. On average, the 4-LSTM layer model took 69% longer training time than the 3-LSTM layer model. However, the validation loss and perplexity were not improved significantly or became even worse for specific conditions. On the other hand, when comparing the automatically generated sentences, the 4-LSTM layer model tended to generate the sentences which are closer to the natural language than the 3-LSTM model. Although there were slight differences in the completeness of the generated sentences between the models, the sentence generation performance was quite satisfactory in any simulation conditions: they generated only legitimate Korean letters and the use of postposition and the conjugation of verbs were almost perfect in the sense of grammar. The results of this study are expected to be widely used for the processing of Korean language in the field of language processing and speech recognition, which are the basis of artificial intelligence systems.

• Journal of Intelligence and Information Systems
• /
• v.26 no.1
• /
• pp.97-117
• /
• 2020

Product evaluation criteria is an indicator describing attributes or values of products, which enable users or manufacturers measure and understand the products. When companies analyze their products or compare them with competitors, appropriate criteria must be selected for objective evaluation. The criteria should show the features of products that consumers considered when they purchased, used and evaluated the products. However, current evaluation criteria do not reflect different consumers' opinion from product to product. Previous studies tried to used online reviews from e-commerce sites that reflect consumer opinions to extract the features and topics of products and use them as evaluation criteria. However, there is still a limit that they produce irrelevant criteria to products due to extracted or improper words are not refined. To overcome this limitation, this research suggests LDA-k-NN model which extracts possible criteria words from online reviews by using LDA and refines them with k-nearest neighbor. Proposed approach starts with preparation phase, which is constructed with 6 steps. At first, it collects review data from e-commerce websites. Most e-commerce websites classify their selling items by high-level, middle-level, and low-level categories. Review data for preparation phase are gathered from each middle-level category and collapsed later, which is to present single high-level category. Next, nouns, adjectives, adverbs, and verbs are extracted from reviews by getting part of speech information using morpheme analysis module. After preprocessing, words per each topic from review are shown with LDA and only nouns in topic words are chosen as potential words for criteria. Then, words are tagged based on possibility of criteria for each middle-level category. Next, every tagged word is vectorized by pre-trained word embedding model. Finally, k-nearest neighbor case-based approach is used to classify each word with tags. After setting up preparation phase, criteria extraction phase is conducted with low-level categories. This phase starts with crawling reviews in the corresponding low-level category. Same preprocessing as preparation phase is conducted using morpheme analysis module and LDA. Possible criteria words are extracted by getting nouns from the data and vectorized by pre-trained word embedding model. Finally, evaluation criteria are extracted by refining possible criteria words using k-nearest neighbor approach and reference proportion of each word in the words set. To evaluate the performance of the proposed model, an experiment was conducted with review on '11st', one of the biggest e-commerce companies in Korea. Review data were from 'Electronics/Digital' section, one of high-level categories in 11st. For performance evaluation of suggested model, three other models were used for comparing with the suggested model; actual criteria of 11st, a model that extracts nouns by morpheme analysis module and refines them according to word frequency, and a model that extracts nouns from LDA topics and refines them by word frequency. The performance evaluation was set to predict evaluation criteria of 10 low-level categories with the suggested model and 3 models above. Criteria words extracted from each model were combined into a single words set and it was used for survey questionnaires. In the survey, respondents chose every item they consider as appropriate criteria for each category. Each model got its score when chosen words were extracted from that model. The suggested model had higher scores than other models in 8 out of 10 low-level categories. By conducting paired t-tests on scores of each model, we confirmed that the suggested model shows better performance in 26 tests out of 30. In addition, the suggested model was the best model in terms of accuracy. This research proposes evaluation criteria extracting method that combines topic extraction using LDA and refinement with k-nearest neighbor approach. This method overcomes the limits of previous dictionary-based models and frequency-based refinement models. This study can contribute to improve review analysis for deriving business insights in e-commerce market.

• Korean Journal of Psychosomatic Medicine
• /
• v.4 no.1
• /
• pp.37-43
• /
• 1996

This study was designed to investigate the relationship between changes in manic symptoms and in thyroid hormone level in manic patients. Thyroid hormones(T3, T4) and TSH were measured, and Young's Mania Rating Scale(YMRS) was scored in twenty manic inpatients, at the time of admission and 6 weeks later. Mean value of T3 was slightly increased(increase : 9, decrease : 11). Mean value of T4 was fairly decreased(increase 4, decrease 16), while that of TSH was increased(increase 14, decrease : 6). Mean total score of YMRS was fairly decreased and mean values of each eleven subscales were decreased. Subjects were classified as increased, decreased and unchanged group for each thyroid hormones. for T3, there was statistically significant differences in YMRS subscale of 6(speech : speed/amount), 7(language, thought disturbance) and 10(appereance) between 3 groups. With the above results, the changes of manic symptom during the course of illness could probably be indicated by the thyroid hormon change, and manic symptoms could be classified as subtypes in terms of thyroid hormone change.

• The Journal of Korean Academy of Sensory Integration
• /
• v.14 no.2
• /
• pp.33-45
• /
• 2016

Objective : The purpose of this article is to study the effectiveness of a group occupational therapy for children via systematic review, and to provide clinical evidences. Methods : The systematic review initially included articles in Pubmed, Ovid, RISS, KISS, and DBpia, published until 2014. Total 14 articles have been chosen, using different keyword of each domestic/foreign studies. Selected articles were divided depending on the level of evidence, and eventually been organized into group PICO (Patient, Intervention, Comparison, Outcome). Results : The results indicated that there are more diagnosed children than normal in participants and especially children with sensory integration disorder (28.6%) appeared to be the most common. In intervention, group sensory integrative intervention had the largest frequency, and instruments measuring social skills were used most for analyzing the intervention results. The effectiveness of interventions showing statistically significant results were 40.7% of Total. Conclusion : Group occupational therapy for children should be provided to various diagnosis with widely better quality of interventions providing the enough support for the evidence level.

• Journal of The Korean Association For Science Education
• /
• v.35 no.3
• /
• pp.419-429
• /
• 2015

In this study, we explored students' epistemological framing during scientific argumentation and how interactions among group members influenced group argumentation. Twenty-one gifted science students divided into groups of three or four participated in this study. Students' discussions related to data interpretation concerning the rate of photosynthesis were analyzed. Students' activities were videotaped in groups so the discourse could be transcribed and students' behavioral cues analyzed. Students' epistemological framing has been identified through analysis of their speech and behavioral responses to the anomalous data from the inquiry process. Subsequently, their sources of warrant and group argumentation levels were explored. We found out that group members framed the inquiry in two ways: "understanding phenomena" and "classroom game." Group members whose framing was "understanding phenomena" required other members to justify the anomalous data by examining its validity and reliability, which conclusively demonstrated a high level of argumentation. On the other hand, when group members used "classroom game" to frame their argumentation, they did not recognize the necessity of explaining the anomalous data; rather, these students used simple empirical justification to explain the data, reflecting a low level of argumentation. When students using different epistemological framing disagreed over interpretations of anomalous data throughout the discussion, clashes ensued that resulted in emotional conflict and a lack of discussion. Students' framing shifts were observed during the discussion on which group leaders seemed to have a huge influence. This study lays the foundation for future work on establishing productive framing to prompt scientific argumentation in science classrooms.

• The Journal of Korean Academy of Prosthodontics
• /
• v.36 no.2
• /
• pp.258-272
• /
• 1998

This investigation evaluated patients who received Steri-Oss implants from the Dental Hospital of Chosun University during the period from March 1989 to August 1997. 346 fixtures of 127 patients were included in this study. The results were as follows ; 1.The follow-up period was defined as the period between the surgical placement of the implants and the last follow-up examination. The mean follow-up period was $2.17{\pm}1.21$ years. 2.The period between fixture installation and second surgery was $0.71{\pm}0.44$ years in the maxilla and $0.46{\pm}0.21$ years in the mandible. 3.The number of fixtures which were installed in the upper jaw(112) was less than that in the lower jaw(234) and in the posterior region(260) was more than in the anterior region(86). 4.The length of fixture which was most frequently used was 12 mm and least was 8mm. Screw implants were installed more than cylindrical implants. 3.8mm implant was the most common implans, followed by 4.5mm and 3.25mm. 5.The number of augmentation cases was more than that of non-augmentation cases and the rate of augmentation cases in the maxilla was more than that in the mandible. 6.Implant restorations for partial edentulos patients(94cases) were more than single- tooth implant restorations(33cases) or implant restorations for complete edentulos patients(10cases). 7.Free-standing prostheses for partially edentulous patients were more commom than any other type of connection between implants and natural teeth. 8.Plaque Index($0.95{\pm}0.74$) and Gingival Index($0.31{\pm}0.52$) were very similar around the natural teeth and reflected an acceptable level of plaque and gingivitis control. Mean value for keratinized mucosa index($1.93{\pm}1.20$) remained fairly constant around level 2(1-2 mm keratinized epithelium). 9.Patients were generally satisfied with implant in terms of comfort, function, speech and esthetics. 10.There was not a statistically significant differences in overall survial rate between implants placed in the maxilla (91.5%) and those placed in the mandible (93.8%). Fourteen implants lost before the prosthetic rehabilitation and eleven implants lost following variable periods in function after the prosthetic phase of the treatment. 11.Cause of implant failures was exfoliation or removal of fixture due to non-osseointegration before the prosthetic rehabilitation or due to fracture of fixture, masticatory pain after the prosthetic rehabilitation. 12.The survival rate of Steri-Oss implants using the Kaplan-Meier statistical analysis was 93.8% at 2 year and 86.6% at 5 year, In all cases, implant losses occured predominantly in the healing period. There was a steep decline in the rate of implant loss after the first year. 13.The survival rate of Steri-Oss implants in the anterior region was 94.8% at 2 year and 94.8% at 5 year and that in the posterior region was 92.8% at 2 year and 75.9% at 5 year. In conclusion, this study revealed a number of parameters and guidelines for achieving an optimal success rate in osseointegration.

• Journal of the Korean Society of Clothing and Textiles
• /
• v.34 no.2
• /
• pp.226-241
• /
• 2010

This study details the history and characteristics of consumption values, text style analyses, and appeal types expressed in magazine commercials from 1955 to 2008. This study analyzes the level of the social structure of commercial expression for each period. Consumption values based on the categories of consumption values by Sheth (1991) were classified through a total commercials analysis. Analyses on closing types of sentences, types of sentences, and rhetorical figures were executed focusing on headline text and text style. Appealing types were composed of rational, emotional, and ethical appeals. For analysis, the crosstab analysis and chi-square test of SPSS are used. The results are as follow. Seven values were constructed, functional value, social value, emotional value, conditional value, epistemic value, fashionable value, and indistinct value. The ratio of emotional value was the highest and functional value, epistemic value conditional value, fashionable value, social value, and indistinct value followed. The emotional value social value, conditional value, fashionable value, and epistemic value that focused on the emotion of consumers increased, while the functional value decreased. Sentences that use narrative styles, hyperboles, and metaphors that increased the interest of readers were dominantly used in the headline texts. For sentence expression, a declarative sentence in a sentence type, exciting curiosity in the expression method where hyperbole and figures of speech in rhetorical expressions are used most often. Emotional appeal was used almost twice more than the reasonable appeal for appeal types of the total commercial. The lower level of reasonable appeal is information that provides the product function. Interest and expression (such as pleasure and achievement) were used most often for emotional appeal. These results show that the most important issue is the emotional value in consumption in understanding the consumer. Marketing managers should also be aware of the functional value as well as an emotional value.

• The Journal of Korean Academy of Prosthodontics
• /
• v.30 no.3
• /
• pp.338-360
• /
• 1992

This study was performed to investigate the phonetic alterations with upper anterior teeth were missing. To compare the changes of the phonations, before and after insertion of the temporary prosthesis, six subjects who lost their upper anterior teeth were selected (2-male, 4-female). Tested sounds (/ga(가), na(나), da(다), ra(라), sa(사), ja(자), cha(차), ta(타), pa(파), ha(하), gi(기), ni(니), di(디), ri(리), si(시), jl(지), chi(치), ti(티), pi(피), hi(히), seu(스), se(세), so(소), su(수)/were programmed into an IBM AT with and without temporary prosthesis. These experiments were analyzed by formants, consonants durations, and energy level changes with an LSI speech work station program. During the pronunciation of the tested sounds (with and without temporary prosthesis), mandibular movements were recorded to a Mandibular Kinesiogram and analyzed . The findings led to the following conclusions: 1. Objective differences could not be found. However, in every informant, subjective improvement could be noticed. 2. There were no persistant correlations of the formant's changes. And in every informant, phonetic changes were variable. 3. There were various changes of the consonant durations in every informant. By and large, those of /si(시), jl(지), chi(치), Pi(피), hi(히)/ were longer than other tested sounds. After insertion of the prosthesis, durations were shorter. Consonants with /i(ㅣ)/ were longer than with /a(ㅏ)/, with or without prosthesis. 4. With and without temporary prosthesis, mandibular movements were various in the frontal view. Mandibular movements showed lateral deviations, and mandibular positions with /si(시), ji(지), ti(티), seu(스), hi(히)/ were nearer to the mandibular rest position. 5. The kinds of temporary prosthesis and conditions of the missing teeth influenced every informant variously, so there were no correlation between informants. 6. Energy levels increased in all tested sounds with a fixed temporary prosthesis. And, there were no differences between before and after insertion of a removable temporary prosthesis. However, sibilant sounds, and consonants with /i(ㅣ)/ showed a little increased energy level.