• Proceedings of the Acoustical Society of Korea Conference
• /
• autumn
• /
• pp.81-84
• /
• 1999

본 논문에서는 중규모급 단어인식기의 실시간 구현을 위한 무감독 단어집단화 알고리듬을 제안한다. 무감독 단어집단화는 인식대상 어휘 수가 많은 대용량 음성인식 시스템에서 대상 어휘 수를 줄여주는 역할을 하는 전처리기의 성격을 갖는다. 무감독 집단화를 위해 각 단어의 유$\cdot$무성음 고유의 특성을 잘 반영할 수 있는 특징 파라미터 5개를 사용하여 패턴 인식과 회귀분석에서 널리 사용되고 있는 분류$\cdot$회귀트리(Classification And Regression Tree)에 적용시키는 방법으로 접근하였고, 각 단어의 frame 수를 일정하게 n개로 분할(segment)하여 1개의 tree를 생성시키는 방법과 각 segment에 해당하는 tree를 생성시켜 segment들 사이의 교집합 성분으로 단어들을 집단화 하였다 실험결과 탐색 대상단어 22개에서 평균2.21개로 줄어 전체 대상 단어의 $10\%$만을 탐색하여 인식할 수 있는 방법을 제시할 수 있었다.

• Journal of Korean Society for Quality Management
• /
• v.48 no.3
• /
• pp.421-431
• /
• 2020

Purpose: The purpose of this study is to compare and analyze the tokenizer in natural language processing for customer satisfaction in sentiment analysis. Methods: In this study, a supervised learning-based tokenizer Mecab-Ko and an unsupervised learning-based tokenizer SentencePiece were used for comparison. Three algorithms: Naïve Bayes, k-Nearest Neighbor, and Decision Tree were selected to compare the performance of each tokenizer. For performance comparison, three metrics: accuracy, precision, and recall were used in the study. Results: The results of this study are as follows; Through performance evaluation and verification, it was confirmed that SentencePiece shows better classification performance than Mecab-Ko. In order to confirm the robustness of the derived results, independent t-tests were conducted on the evaluation results for the two types of the tokenizer. As a result of the study, it was confirmed that the classification performance of the SentencePiece tokenizer was high in the k-Nearest Neighbor and Decision Tree algorithms. In addition, the Decision Tree showed slightly higher accuracy among the three classification algorithms. Conclusion: The SentencePiece tokenizer can be used to classify and interpret customer sentiment based on online reviews in Korean more accurately. In addition, it seems that it is possible to give a specific meaning to a short word or a jargon, which is often used by users when evaluating products but is not defined in advance.

• The Journal of Korean Association of Computer Education
• /
• v.11 no.3
• /
• pp.67-80
• /
• 2008

Various document categorization methods have been studied to provide a user with an effective way of browsing a large scale of documents. They do compares set of documents into groups of semantically similar documents automatically. However, the automatic categorization method suffers from low accuracy. This thesis proposes a semi-automatic document categorization method based on the domains of documents. Each documents is belongs to its initial domain. All the documents in each domain are recursively clustered in a level-wise manner, so that the category tree of the documents can be founded. To find the clusters of documents, the stop-word of each document is removed on the document frequency of a word in the domain. For each cluster, its cluster keywords are extracted based on the common keywords among the documents, and are used as the category of the domain. Recursively, each cluster is regarded as a specified domain and the same procedure is repeated until it is terminated by a user. In each level of clustering, a user can adjust any incorrectly clustered documents to improve the accuracy of the document categorization.

• Journal of Korean Elementary Science Education
• /
• v.35 no.2
• /
• pp.150-165
• /
• 2016

This study aims to analyze types of science museum worksheets developed by elementary pre-service teachers and their perspectives on the requirements and necessity of science museum worksheets. As analysis subjects, this study selected 38 kinds of worksheets and reports developed by 114 elementary pre-service teachers who were in the third year of university of education. In this study, the science museum selected for elementary pre-service teachers to develop worksheets was a national science museum, composed of 'Nature and Discovery Museum', 'Science Technology and Industry Museum' and 'Children's Museum', which was located in a metropolitan city and opened in 2013. The results of this study can be summarized as follows; Firstly, as a result of analyzing the science museum worksheets developed by elementary pre-service teachers, this study found out that the experience type with hands-on and observation techniques applied was most, and as an approach method, direct manipulation, look-in observation and close observation were most. However, although these science museum worksheets were experience-oriented, many of them were survey-oriented ones that suggested too many questions through various exhibits. Secondly, as a result of analyzing requirements of science museum worksheets elementary pre-service teachers thought and described through the word tree of NVivo 10, this study extracted 10 kinds of main themes, out of which the requirement, 'A limited amount of activity should be required', showed the highest frequency. Thirdly, as a result of analyzing the necessity of science museum worksheets elementary pre-service teachers thought and described through the word tree of NVivo 10, this study extracted 9 kinds of main themes, out of which the opinion, 'It is required to help students check an exhibit which may be passed by', was most.

• The Journal of the Korea Contents Association
• /
• v.20 no.6
• /
• pp.556-578
• /
• 2020

There are many different researches about how to analyze issues based on real-time news streams. But, there are few researches which analyze issues hierarchically from news articles and even a previous research of hierarchical issue analysis make clustering speed slower as the increment of news articles. In this paper, we propose a hierarchical and incremental clustering for semi real-time issue analysis on news articles. We trained siamese neural network based weighted cosine similarity model, applied this model to k-means algorithm which is used to make word clusters and converted news articles to document vectors by using these word clusters. Finally, we initialized an issue cluster tree from document vectors, updated this tree whenever news articles happen, and analyzed issues in semi real-time. Through the experiment and evaluation, we showed that up to about 0.26 performance has been improved in terms of NMI. Also, in terms of speed of incremental clustering, we also showed about 10 times faster than before.

• The Korean Journal of Applied Statistics
• /
• v.29 no.7
• /
• pp.1257-1270
• /
• 2016

This study deals with the prediction of the total number of movie audiences as a measure for the box office. Prediction is performed by classification techniques of data mining such as decision tree, multilayer perceptron(MLP) neural network model, multinomial logit model, and support vector machine over time such as before movie release, release day, after release one week, and after release two weeks. Predictors used are: online word-of-mouth(OWOM) variables such as the portal movie rating, the number of the portal movie rater, and blog; in addition, other variables include showing the inherent properties of the film (such as nationality, grade, release month, release season, directors, actors, distributors, the number of audiences, and screens). When using 10-fold cross validation technique, the accuracy of the neural network model showed more than 90 % higher predictability before movie release. In addition, it can be seen that the accuracy of the prediction increases by adding estimates of the final OWOM variables as predictors.

• The Journal of the Acoustical Society of Korea
• /
• v.18 no.6
• /
• pp.49-56
• /
• 1999

In many continuous speech recognition systems based on HMMs, decision tree-based state tying has been used for not only improving the robustness and accuracy of context dependent acoustic modeling but also synthesizing unseen models. To construct the phonetic decision tree, standard method performs one-level pruning using just single Gaussian triphone models. In this paper, two novel approaches, two-level decision tree and multi-mixture decision tree, are proposed to get better performance through more accurate acoustic modeling. Two-level decision tree performs two level pruning for the state tying and the mixture weight tying. Using the second level, the tied states can have different mixture weights based on the similarities in their phonetic contexts. In the second approach, phonetic decision tree continues to be updated with training sequence, mixture splitting and re-estimation. Multi-mixture Gaussian as well as single Gaussian models are used to construct the multi-mixture decision tree. Continuous speech recognition experiment using these approaches on BN-96 and WSJ5k data showed a reduction in word error rate comparing to the standard decision tree based system given similar number of tied states.

• Annual Conference on Human and Language Technology
• /
• 1998.10c
• /
• pp.419-424
• /
• 1998

본 논문에서는 단어들의 분포적 특성을 이용하여 자동으로 단어를 군집화(clustering) 하는 기법을 제시한다. 제안된 군집화 기법에서는 단어들 사이의 거리(distance)를 가상 공간상에 있는 두 단어의 평균점에 대한 불일치의 합(total divergence to the average)으로 측정하며 군집화 알고리즘으로는 최소 신장 트리(minimal spanning tree)를 이용한다. 본 논문에서는 이 기법에 대해 두 가지 실험을 수행한다. 첫 번째 실험은 코퍼스에서 상위 출현 빈도를 가지는 약 1,200 개의 명사들을 의미에 따라 군집화 하는 것이며 두 번째 실험은 이 논문에서 제시한 자동 군집화 방법의 성능을 객관적으로 평가하기 위한 것으로 가상 단어(pseudo word)에 대한 군집화이다. 실험 결과 이 방법은 가상 단어에 대해 약 91%의 군집화 정확도와(clustering precision)와 약 81%의 군집 순수도(cluster purity)를 나타내었다. 한편 두 번째 실험에서는 평균점에 대한 불일치의 합을 이용한 거리 측정에서 나타나는 문제점을 보완한 거리 측정 방법을 제시하였으며 이를 이용하여 가상 단어 군집화를 수행한 결과 군집화 정확도와 군집 순수도가 각각 약 96% 및 95%로 향상되었다.

• Journal of KIISE:Software and Applications
• /
• v.31 no.4
• /
• pp.488-497
• /
• 2004

This paper describes the hybrid document summarization using the indicative summarization and the query-based summarization. The learning models are built from teaming documents in order to extract topic phrases. We use Naive Bayesian, Decision Tree and Supported Vector Machine as the machine learning algorithm. The system extracts topic phrases automatically from new document based on these models and outputs the summary of the document using query-based summarization which considers the extracted topic phrases as queries and calculates the locality-based similarity of each topic phrase. We examine how the topic phrases affect the summarization and how many phrases are proper to summarization. Then, we evaluate the extracted summary by comparing with manual summary, and we also compare our summarization system with summarization mettled from MS-Word.

• The Korean Journal of Applied Statistics
• /
• v.29 no.4
• /
• pp.657-678
• /
• 2016

This study deals with the effect of online word-of-mouth (OWOM) variables on the box office. From the result of statistical analysis on 276 films with audiences of more than five hundred thousand released in the Korea from 2012 to 2015, it can be seen that the variables showing the size of OWOM (such as the number of the portal movie rater, blog, and news after release) are associated more with the box office than the portal movie rating showing the direction of OWOM as well as variables showing the inherent properties of the film such as grade, nationality, release month, release season, directors, actors, and distributors.