TOPIK Automatic Writing Assessment
Research and development of an automated scoring system for the TOPIK (Test of Proficiency in Korean) writing section, with multi-trait scoring and rationale generation.
Period: Feb. 2024 — Oct. 2024
Organization: NSDevil Inc., National Institute for International Education (NIIED)
As the principal investigator, I developed an automatic scoring model for the TOPIK writing section. I enhanced the model explainability by multi-trait scoring and generating rationales for the model-rated scores.
TOPIK (Test of Proficiency in Korean) is an official exam to assess the Korean language proficiency of non-native speakers, administered by the Korean national institution.