본문 바로가기
  • Home

Deep Learning-Based Classification of Special Education Logs Using DistilKoBERT

  • Journal of The Korea Society of Computer and Information
  • Abbr : JKSCI
  • 2025, 30(6), pp.55~64
  • Publisher : The Korean Society Of Computer And Information
  • Research Area : Engineering > Computer Science
  • Received : April 15, 2025
  • Accepted : June 4, 2025
  • Published : June 30, 2025

Won-Cheol Park 1

1국립공주대학교

Accredited

ABSTRACT

This study proposes a DistilKoBERT-based model that automatically classifies disability types by reflecting the learning characteristics of special education students. The dataset consists of 3,425 sentences extracted from special education support logs and is categorized into five disability types: Intellectual Disability, Emotional and Behavioral Disorders, Communication Disorders, Autism Spectrum Disorder, and Physical Disability. Text mining and natural language processing techniques were applied for data preprocessing. Additionally, the performance of the DistilKoBERT model was compared with TF-IDF + SVM and XGBoost. The evaluation results demonstrated that the DistilKoBERT model achieved the highest accuracy (0.89) and F1-score (0.87). This confirms that the proposed automatic classification system outperforms existing methods. This study highlights the potential of automated diagnostic systems in the field of special education. Future research aims to enhance the system’s practicality by expanding the dataset and optimizing the model.

Citation status

* References for papers published after 2023 are currently being built.