Proceedings of the 3rd International Conference on Educational Innovation and Multimedia Technology, EIMT 2024, March 29–31, 2024, Wuhan, China

Research Article

Programming Knowledge Tracing based on Problem Solution Embedding

Download51 downloads
  • @INPROCEEDINGS{10.4108/eai.29-3-2024.2347724,
        author={Yongfeng  Huang and Rongfang  Wang},
        title={Programming Knowledge Tracing based on Problem Solution Embedding},
        proceedings={Proceedings of the 3rd International Conference on Educational Innovation and Multimedia Technology, EIMT 2024, March 29--31, 2024, Wuhan, China},
        publisher={EAI},
        proceedings_a={EIMT},
        year={2024},
        month={6},
        keywords={knowledge tracing; performance prediction; feature fusion; codebert},
        doi={10.4108/eai.29-3-2024.2347724}
    }
    
  • Yongfeng Huang
    Rongfang Wang
    Year: 2024
    Programming Knowledge Tracing based on Problem Solution Embedding
    EIMT
    EAI
    DOI: 10.4108/eai.29-3-2024.2347724
Yongfeng Huang1,*, Rongfang Wang1
  • 1: Donghua University
*Contact email: yfhuang@dhu.edu.cn

Abstract

With the development of the internet and information technology, using online learning systems for programming practice has increasingly become a new trend. In this process, continuously tracking students' proficiency in programming skills is also particularly important. However, current research on students' programming practice mainly focuses on their submitted code, neglecting the importance of official exercise solutions in programming competitions for assessing students' proficiency in programming skills. Therefore, we propose a new improved model for graph-based knowledge tracing(CTGKT). Specifically, by embedding official exercise solutions, combined with students' submitted codes and the exercise text contents , it assesses the proficiency of students in programming skills. Experiments on our programming competition practice dataset demonstrate that CTGKT model achieves state-of-the-art performance compared to existing methods.