Fine-tuned KoR-SRoBERTa for Corporate News Relevance Classification
⸻
Overview
This model is a fine-tuned version of the base model jhgan/ko-sroberta-multitask, adapted specifically for relevance classification of Korean corporate-related news articles. The goal of this model is to improve performance in distinguishing significant news within firm-specific news contexts, which often contain domain-specific financial language.
This work is based on the methodology and dataset presented in the following academic paper:
Hyun Ji-won, Lee Jun-il, and Cho Hyun-kwon (2022).
“A Study on Sentiment Classification of Corporate-related News Articles Using KoBERT.”
Accounting Research 47(4), 33–54.
https://www.kci.go.kr/kciportal/ci/sereArticleSearch/ciSereArtiView.kci?sereArticleSearchBean.artiId=ART002873618This model card includes proper attribution to the original authors of the base model as required under the CC-BY-SA-4.0 license.
⸻
Model Details
• Base model: jhgan/ko-sroberta-multitask
• Architecture: RoBERTa (Korean SRoBERTa variant)
• Task: Relevance classification (1/0)
• Language: Korean
• Domain: Corporate & financial news (Korean)
• Fine-tuning: Conducted on a curated dataset of company-related news headlines and lead sentences derived from the methodology of the 2022 study.
⸻
Intended Use
This model is intended for:
• Identifying relevance of firm-specific and finance-related Korean news
• Downstream applications requiring corporate textual analytsis
Not intended for:
• General-purpose Korean analysis without domain adaptation
• High-stakes financial decision-making without human oversight
⸻
Licensing
This model is published under CC-BY-SA-4.0, inherited from the base model:
• Original base model: jhgan/ko-sroberta-multitask
• Original license: CC-BY-SA-4.0
⸻
Attribution
If you use this model, please cite both:
- Base Model
jhgan/ko-sroberta-multitask
(Original authors as listed on the model card)
- Fine-tuned Model / Academic Basis
Hyun Ji-won, Lee Jun-il, and Cho Hyun-kwon (2022).
A Study on Sentiment Classification of Corporate-related News Articles Using KoBERT.
Accounting Research 47(4), 33–54.
https://www.kci.go.kr/kciportal/ci/sereArticleSearch/ciSereArtiView.kci?sereArticleSearchBean.artiId=ART002873618
