As pre-trained language models become more widely used, the demand for datasets tailored to specialized domains has grown. To meet this need in the environmental, social, and governance (ESG) space, researchers have constructed ESG-Kor, a Korean-language dataset focused on ESG-related information extraction.
The dataset is designed to support tasks such as identifying and extracting ESG-relevant information from text, which could aid in automating ESG assessments and reporting. While the abstract does not provide details on the dataset's size or annotation methodology, it emphasizes practical use cases, suggesting the resource is intended for real-world applications.
This work highlights a broader trend of building domain-specific datasets for languages beyond English, enabling more accurate and context-aware language model performance in specialized fields like ESG analysis.