Pre-training DataText

Game Articles Raw Dataset (Press Release · News · Reporting · Interview)

Type
Document Dataset
Domain
Games
Language
Korean

Overview

This dataset is a Korean raw-text corpus collected from game media, including publisher press releases, news updates, event coverage, and developer interviews.

Potential Use Cases

  • Domain Pre-Training for Korean Gaming Text: Supplies the title, studio, genre, and live-service vocabulary that general Korean corpora underrepresent, adapting models to the gaming domain.
  • Announcement vs. Reporting Discrimination: The presence of both promotional and editorial text on the same subjects supports training classifiers that distinguish announcement framing from independent assessment.
Flitto Curation Data Flitto Curation Data