• Speech Sciences
• /
• v.8 no.1
• /
• pp.77-91
• /
• 2001

This paper investigates the timing of Korean spoken in a news-reading speech style in order to improve the naturalness of durations used in Korean speech synthesis. Each segment in a corpus of 671 read sentences was annotated with 69 segmental and prosodic features so that the measured duration could be correlated with the context in which it occurred. A CART model based on the features showed a correlation coefficient of 0.79 with an RMSE (root mean squared prediction error) of 23 ms between actual and predicted durations in reserved test data. These results are comparable with recent published results in Korean and similar to results found in other languages. An analysis of the classification tree shows that phrasal structure has the greatest effect on the segment duration, followed by syllable structure and the manner features of surrounding segments. The place features of surrounding segments only have small effects. The model has application in Korean speech synthesis systems.

• Journal of KIISE:Software and Applications
• /
• v.41 no.8
• /
• pp.580-585
• /
• 2014

This paper conducts a variety of experiments for "the extraction of Korean parallel sentences using Wikipedia data". We refer to various methods that were previously proposed for other languages. We use two approaches. The first one is to use translation probabilities that are extracted from the existing resources such as Sejong parallel corpus, and the second one is to use dictionaries such as Wiki dictionary consisting of Wikipedia titles and MRDs (machine readable dictionaries). Experimental results show that we obtained a significant improvement in system using Wikipedia data in comparison to one using only the existing resources. We finally achieve an outstanding performance, an F1-score of 57.6%. We additionally conduct experiments using a topic model. Although this experiment shows a relatively lower performance, an F1-score of 51.6%, it is expected to be worthy of further studies.

• International Journal of Computer Science & Network Security
• /
• v.22 no.9
• /
• pp.334-342
• /
• 2022

Nowadays, as we can notice on social media, most users choose to use more than one language in their online postings. Thus, social media analytics needs reviewing as code-switching analytics instead of traditional analytics. This paper aims to present evidence comparable to the accuracy of code-switching analytics techniques in analysing the mood state of social media users. We conducted a systematic literature review (SLR) to study the social media analytics that examined the effectiveness of code-switching analytics techniques. One primary question and three sub-questions have been raised for this purpose. The study investigates the computational models used to detect and measures emotional well-being. The study primarily focuses on online postings text, including the extended text analysis, analysing and predicting using past experiences, and classifying the mood upon analysis. We used thirty-two (32) papers for our evidence synthesis and identified four main task classifications that can be used potentially in code-switching analytics. The tasks include determining analytics algorithms, classification techniques, mood classes, and analytics flow. Results showed that CNN-BiLSTM was the machine learning algorithm that affected code-switching analytics accuracy the most with 83.21%. In addition, the analytics accuracy when using the code-mixing emotion corpus could enhance by about 20% compared to when performing with one language. Our meta-analyses showed that code-mixing emotion corpus was effective in improving the mood analytics accuracy level. This SLR result has pointed to two apparent gaps in the research field: i) lack of studies that focus on Malay-English code-mixing analytics and ii) lack of studies investigating various mood classes via the code-mixing approach.

• Phonetics and Speech Sciences
• /
• v.8 no.2
• /
• pp.11-16
• /
• 2016

This study explores the phoneme distribution and syllable structure of entry words in the CMU English Pronouncing Dictionary to provide phoneticians and linguists with fundamental phonetic data on English word components. Entry words in the dictionary file were syllabified using an R script and examined to obtain the following results: First, English words preferred consonants to vowels in their word components. In addition, monophthongs occurred much more frequently than diphthongs. When all consonants were categorized by manner and place, the distribution indicated the frequency order of stops, fricatives, and nasals according to manner and that of alveolars, bilabials and velars according to place. These results were comparable to the results obtained from the Buckeye Corpus (Yang, 2012). Second, from the analysis of syllable structure, two-syllable words were most favored, followed by three- and one-syllable words. Of the words in the dictionary, 92.7% consisted of one, two or three syllables. This result may be related to human memory or decoding time. Third, the English words tended to exhibit discord between onset and coda consonants and between adjacent vowels. Dissimilarity between the last onset and the first coda was found in 93.3% of the syllables, while 91.6% of the adjacent vowels were different. From the results above, the author concludes that an analysis of the phonetic symbols in a dictionary may lead to a deeper understanding of English word structures and components.

• Journal of Advanced Marine Engineering and Technology
• /
• v.39 no.3
• /
• pp.274-279
• /
• 2015

The main goal of this research is to build a sentiment analysis system which automatically determines user opinions of the Stanford Sentiment Treebank in terms of three sentiments such as positive, negative, and neutral. Firstly, sentiment sentences are POS tagged and parsed to dependency structures. All nodes of the Treebank and their polarities are automatically extracted from the Treebank. We train two Support Vector Machines models. One is for a node level classification and the other is for a sentence level. We have tried various type of features such as word lexicons, POS tags, Sentiment lexicons, head-modifier relations, and sibling relations. Though we acquired 74.2% in accuracy on the test set for 3 class node level classification and 67.0% for 3 class sentence level classification, our experimental results for 2 class classification are comparable to those of the state of art system using the same corpus.

• The Journal of the Acoustical Society of Korea
• /
• v.23 no.8
• /
• pp.583-589
• /
• 2004

Proposed is a speaker normalization method based on vector quantizer for continuous speech recognition (CSR) system in which no acoustic information is made use of. The proposed method, which is an improvement of the previously reported speaker normalization scheme for a simple digit recognizer, builds up a canonical codebook by iteratively training the codebook while the size of codebook is increased after each iteration from a relatively small initial size. Once the codebook established, the warp factors of speakers are estimated by comparing exhaustively the warped versions of each speaker's utterance with the codebook. Two sets of phones are used to estimate the warp factors: one, a set of vowels only. and the other, a set composed of all the Phonemes. A Piecewise linear warping function which corresponds to the estimated warp factor is adopted to warp the power spectrum of the utterance. Then the warped feature vectors are extracted to be used to train and to test the speech recognizer. The effectiveness of the proposed method is investigated by a set of recognition experiments using the TIMIT corpus and HTK speech recognition tool kit. The experimental results showed comparable recognition rate improvement with the formant based warping method.

• Korean Journal of Cognitive Science
• /
• v.14 no.1
• /
• pp.29-40
• /
• 2003

Cross-Language Information Retrieval (CLIR) deals with the documents in various languages by one language query. A user who uses one language can retrieve the documents in another language through CLIR system. In CLIR, query translation method is known to be more efficient. For the better performance of query translation, we need more resources like dictionary, ontology, and parallel/comparable corpus but usually not available. This paper proposes a new concept called the Common Base Axis which is adapted to Korean-English Query translation ann a new weighting method in dictionary based query translation. The essential idea is that we can express Korean and English word in one vector space by Common Base Axis and use it in calculating sense distance for query weighting. The experiments show that Common Base Axis gives us good performance without ontology and is especially good for one word query translation.

• Journal of Korean Neurosurgical Society
• /
• v.62 no.3
• /
• pp.353-360
• /
• 2019

Epilepsy surgery that eliminates the epileptogenic focus or disconnects the epileptic network has the potential to significantly improve seizure control in patients with medically intractable epilepsy. Magnetic resonance-guided laser interstitial thermal therapy (MRgLITT) has been an established option for epilepsy surgery since the US Food and Drug Administration cleared the use of MRgLITT in neurosurgery in 2007. MRgLITT is an ablative stereotactic procedure utilizing heat that is converted from laser energy, and the temperature of the tissue is monitored in real-time by MR thermography. Real-time quantitative thermal monitoring enables titration of laser energy for cellular injury, and it also estimates the extent of tissue damage. MRgLITT is applicable for lesion ablation in cases that the epileptogenic foci are localized and/or deep-seated such as in the mesial temporal lobe epilepsy and hypothalamic hamartoma. Seizure-free outcomes after MRgLITT are comparable to those of open surgery in well-selected patients such as those with mesial temporal sclerosis. Particularly in patients with hypothalamic hamartoma. In addition, MRgLITT can also be applied to ablate multiple discrete lesions of focal cortical dysplasia and tuberous sclerosis complex without the need for multiple craniotomies, as well as disconnection surgery such as corpus callosotomy. Careful planning of the target, the optimal trajectory of the laser probe, and the appropriate parameters for energy delivery are paramount to improve the seizure outcome and to reduce the complication caused by the thermal damage to the surrounding critical structures.

• Journal of Intelligence and Information Systems
• /
• v.28 no.2
• /
• pp.191-206
• /
• 2022

Recently, it is a de-facto approach to utilize a pre-trained language model(PLM) to achieve the state-of-the-art performance for various natural language tasks(called downstream tasks) such as sentiment analysis and question answering. However, similar to any other machine learning method, PLM tends to depend on the data distribution seen during the training phase and shows worse performance on the unseen (Out-of-Distribution) domain. Due to the aforementioned reason, there have been many efforts to develop domain-specified PLM for various fields such as medical and legal industries. In this paper, we discuss the training of a finance domain-specified PLM for the Korean language and its applications. Our finance domain-specified PLM, KB-BERT, is trained on a carefully curated financial corpus that includes domain-specific documents such as financial reports. We provide extensive performance evaluation results on three natural language tasks, topic classification, sentiment analysis, and question answering. Compared to the state-of-the-art Korean PLM models such as KoELECTRA and KLUE-RoBERTa, KB-BERT shows comparable performance on general datasets based on common corpora like Wikipedia and news articles. Moreover, KB-BERT outperforms compared models on finance domain datasets that require finance-specific knowledge to solve given problems.