• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.510-513
• /
• 2018

본 논문은 다양한 워드 임베딩 모델(word embedding model)들과 하이퍼 파라미터(hyper parameter)들을 조합하였을 때 특정 영역에 어떠한 성능을 보여주는지에 대한 연구이다. 3 가지의 워드 임베딩 모델인 Word2Vec, FastText, Glove의 차원(dimension)과 윈도우 사이즈(window size), 최소 횟수(min count)를 각기 달리하여 총 36개의 임베딩 벡터(embedding vector)를 만들었다. 각 임베딩 벡터를 Fast and Accurate Dependency Parser 모델에 적용하여 각 모들의 성능을 측정하였다. 모든 모델에서 차원이 높을수록 성능이 개선되었으며, FastText가 대부분의 경우에서 높은 성능을 내는 것을 알 수 있었다.

• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.449-452
• /
• 2018

TTS(Text-to-Speech)는 문자열을 입력받아 그 문자열을 음성으로 변환하는 음성합성 기술이다. 그러나 실제 입력되는 문장에는 한글뿐만 아니라 영단어 및 숫자 등이 혼합되어 있다. 영단어는 대소문자에 따라 다르게 읽을 수 있으며, 단위로 사용될 때는 약어로 사용되는 것이므로, 알파벳 단위로 읽어서는 안 된다. 숫자 또한 함께 사용되는 단어에 따라 읽는 방식이 달라진다. 본 논문에서는 한글과 숫자 및 단위, 영단어가 혼합된 문장을 분류하고 이를 음역하는 시스템을 구성하며 word vector를 이용한 숫자 및 단위의 모호성 해소방법을 소개한다.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.4
• /
• pp.1150-1159
• /
• 2000

This paper presents an automatic text categorization model that improves the accuracy by combining statistical and knowledge-based categorization methods. In our model we apply knowledge-based method first, and then apply statistical method on the text which are not categorized by knowledge-based method. By using this combined method, we can improve the accuracy of categorization while categorize all the texts without failure. For statistical categorization, the vector model with Inverted Category Frequency (ICF) weighting is used. For knowledge-based categorization, Phrasal Patterns and Keyword Sets are introduced to represent sentence patterns, and then pattern matching is performed. Experimental results on new articles show that the accuracy of categorization can be improved by combining the tow different categorization methods.

• International Journal of Knowledge Content Development & Technology
• /
• v.7 no.1
• /
• pp.101-120
• /
• 2017

Healthcare professionals have edited many health questions (HQs) and their answers for healthcare consumers on the Internet. The HQs provide both readable and reliable health information, and hence retrieval of those HQs that are relevant to a given question is essential for health education and promotion through the Internet. However, retrieval of relevant HQs needs to be based on the recognition of the intention of each HQ, which is difficult to be done by predefining syntactic and semantic rules. We thus model the intention recognition problem as a text classification problem, and develop two techniques to improve a learning-based text classifier for the problem. The two techniques improve the classifier by location-based and area-based feature weightings, respectively. Experimental results show that, the two techniques can work together to significantly improve a Support Vector Machine classifier in both the recognition of HQ intentions and the retrieval of relevant HQs.

• Proceedings of the Zoological Society Korea Conference
• /
• 2000.10a
• /
• pp.254.1-254
• /
• 2000

No Abstract, See Full Text

• Proceedings of the Korean Magnestics Society Conference
• /
• 2001.04a
• /
• pp.28-29
• /
• 2001

No Abstract, See Full-text

• Proceedings of the Membrane Society of Korea Conference
• /
• 2000.11a
• /
• pp.102-103
• /
• 2000

No Abstract, See Full Text

• Proceedings of the Zoological Society Korea Conference
• /
• 1999.10b
• /
• pp.290.1-290
• /
• 1999

No Abstract, See Full Text

• ETRI Journal
• /
• v.39 no.4
• /
• pp.443-454
• /
• 2017

In this paper, we propose a fine-grained mobile application clustering model using retrofitted document embedding. To automatically determine the clusters and their numbers with no predefined categories, the proposed model initializes the clusters based on title keywords and then merges similar clusters. For improved clustering performance, the proposed model distinguishes between an accurate clustering step with titles and an expansive clustering step with descriptions. During the accurate clustering step, an automatically tagged set is constructed as a result. This set is utilized to learn a high-performance document vector. During the expansive clustering step, more applications are then classified using this document vector. Experimental results showed that the purity of the proposed model increased by 0.19, and the entropy decreased by 1.18, compared with the K-means algorithm. In addition, the mean average precision improved by more than 0.09 in a comparison with a support vector machine classifier.

• The Journal of the Acoustical Society of Korea
• /
• v.20 no.4E
• /
• pp.31-38
• /
• 2001

A large-scale synthesis database for a unit selection based synthesis method usually retains redundant synthesis unit instances, which are useless to the synthetic speech quality. In this paper, to eliminate those instances from the synthesis database, we proposed a new pruning method called weighted vector quantization (WVQ). The WVQ reflects relative importance of each synthesis unit instance when clustering the similar instances using vector quantization (VQ) technique. The proposed method was compared with two conventional pruning methods through the objective and subjective evaluations of the synthetic speech quality: one to simply limit maximum number of instance, and the other based on normal VQ-based clustering. The proposed method showed the best performance under 50% reduction rates. Over 50% of reduction rates, the synthetic speech quality is not seriously but perceptibly degraded. Using the proposed method, the synthesis database can be efficiently reduced without serious degradation of the synthetic speech quality.