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=ART002873618

This 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:

1. Base Model

jhgan/ko-sroberta-multitask
(Original authors as listed on the model card)

2. 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