A Roasted Chestnut Cultivar ‘Jangwon’ with High Pellicle Removability

Park, Yunmi; Kim, Mahn-Jo

doi:10.9787/kjbs.2015.47.4.409

Cited by 5 publications

(13 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the foundation models, we utilize three popular BERT-based models: KoBERT (Lee et al, 2020), KoELECTRA (Park, 2020), and KoBigBird (Park and Kim, 2021). For comparing various model architectures, we have also trained the single-task models that solely classify just one task among the political orientation classification or the level of pro-government classification.…”

Section: Proposed Architecturementioning

confidence: 99%

“…We split the dataset in an 8:1:1 ratio of training, validation, and test datasets. Although the training dataset is unbalanced, we compose a test dataset that has a nearly uniform class distribution imbalance on Ko-KoBERT (Lee et al, 2020), KoBigBird (Park and Kim, 2021), and KoELECTRA (Park, 2020). We use these models from the Hugging Face which provides various Natural Language Processing (NLP) models and datasets.…”

Section: Experiments Setupmentioning

confidence: 99%

“…We conduct the tokenization using the tokenization tool in Hugging Face for each model. In particular, the KoBigBird model (Park and Kim, 2021) can comprehensively process long tokens whose length is more than 2,048 because the model has a relatively large capacity and is devised to properly process long texts. Therefore, KoBigBird (Park and Kim, 2021) is appropriate for analyzing articles of our dataset with tokens above 2048.…”

Section: Experiments Setupmentioning

confidence: 99%

“…We note that our proposed dataset provides a lot of text data that contains diverse political information. To validate the usefulness of our proposed dataset, we leverage state-of-the-art transformer architectures, specifically KoBERT (Lee et al, 2020), KoBigBird (Park and Kim, 2021), and KoELEC-TRA (Park, 2020). We fine-tune the pre-trained models on our proposed dataset to solve the two different classification problems.…”

Section: Introductionmentioning

confidence: 99%

“…In particular, we propose our multi-task model, KoPolitic, which has been trained on our dataset by leveraging the fine-tuning based on the recent KoBigBird (Park and Kim, 2021) architecture. Our proposed text classification model is capable of simultaneously classifying political intent and the level of pro-government.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

The Significance of the Public Private University Policy

Kim¹

2020

Kyung Hee Law Journal

View full text Add to dashboard Cite

We study the charges and first law of thermodynamics for accelerating, nonrotating black holes with dyonic charges in AdS 4 using the covariant phase space formalism. In order to apply the formalism to these solutions (which are asymptotically locally AdS and admit a non-smooth conformal boundary I ) we make two key improvements: 1) We relax the requirement to impose Dirichlet boundary conditions and demand merely a well-posed variational problem. 2) We keep careful track of the codimension-2 corner term induced by the holographic counterterms, a necessary requirement due to the presence of "cosmic strings" piercing I . Using these improvements we are able to match the holographic Noether charges to the Wald Hamiltonians of the covariant phase space and derive the first law of black hole thermodynamics with the correct "thermodynamic length" terms arising from the strings. We investigate the relationship between the charges imposed by supersymmetry and show that our first law can be consistently applied to various classes of non-supersymmetric solutions for which the cross-sections of the horizon are spindles.

show abstract

Section: Proposed Architecturementioning

confidence: 99%

Section: Experiments Setupmentioning

confidence: 99%

Section: Experiments Setupmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

The Significance of the Public Private University Policy

Kim¹

2020

Kyung Hee Law Journal

View full text Add to dashboard Cite

show abstract

Extracting Fallen Objects on the Road From Accident Reports Using a Natural Language Processing Model-Based Approach

Lee,

Cha,

et al. 2023

IEEE Access

View full text Add to dashboard Cite

Keyword extraction is an effective way to quickly identify key elements in text. It can accelerate the identification of key factors that play a role in accidents when applied to incident report analysis. Our research presents an innovative process for extracting keywords from incident reports with the pre-trained natural language processing models. We utilized fine-tuning techniques to integrate a BiLSTM-CRF with a fully-connected layer and pre-trained natural language models. The process of extracting keyphrases is approached as a task of labeling sequences. To analyze incident reports from Korea, we employ pre-trained models customized for the Korean context, such as KoBERT and KoELECTRA. Our approach is assessed using a range of metrics, including accuracy, area under the curve (AUC), F1-score, slot error rate (SER), and simple matching coefficient (SMC). In contrast to traditional approaches which mainly concentrate on document summarization, our research provides a distinct method tailored to identifying falling objects as the main cause of accidents. Our findings demonstrate that the ELECTRA-based model with a BiLSTM-CRF outperforms other models, achieving an accuracy of 0.943, an AUC of 0.991, and a low SER of 0.075. The F1-score and SMC closely resemble the BERT-based model with a BiLSTM-CRF, with no significant differences observed within the 95% confidence interval. These results underscore the potential of fine-tuning pre-trained models for post-hoc traffic accident analysis. This method offers a swift preliminary step to identify the key factors before human analysis, presenting a multifaceted strategy to enhance road safety and prevent accidents.

show abstract

“Why do I feel offended?” - Korean Dataset for Offensive Language Identification

Park,

Kim,

Lee

et al. 2023

Findings of the Association for Computational Linguistics: EACL 2023

View full text Add to dashboard Cite

Warning: This paper contains some offensive expressions.Offensive content is an unavoidable issue on social media. Most existing offensive language identification methods rely on the compilation of labeled datasets. However, existing methods rarely consider low-resource languages that have relatively less data available for training (e.g., Korean). To address these issues, we construct a novel KOrean Dataset for Offensive Language Identification (KODOLI). KODOLI comprises more fine-grained offensiveness categories (i.e., not offensive, likely offensive, and offensive) than existing ones. A likely offensive language refers to texts with implicit offensiveness or abusive language without offensive intentions. In addition, we propose two auxiliary tasks to help identify offensive languages: abusive language detection and sentiment analysis. We provide experimental results for baselines on KODOLI and observe that pre-trained language models suffer from identifying "LIKELY" offensive statements. Quantitative results and qualitative analysis demonstrate that jointly learning offensive language, abusive language and sentiment information improves the performance of offensive language identification.

show abstract

A Roasted Chestnut Cultivar ‘Jangwon’ with High Pellicle Removability

Cited by 5 publications

References 0 publications

The Significance of the Public Private University Policy

The Significance of the Public Private University Policy

Extracting Fallen Objects on the Road From Accident Reports Using a Natural Language Processing Model-Based Approach

“Why do I feel offended?” - Korean Dataset for Offensive Language Identification

Contact Info

Product

Resources

About