Scholarlink

As a crucial practical platform for programming education, Online Judge (OJ) systems directly reflect students’ knowledge mastery and comprehension through their problem-solving performance. Existing evaluations primarily focus on pass/fail status and ranking statistics, while overlooking the learning insights embedded in behavioral data such as submission timing, attempt frequency, and keystroke behaviors. To address this gap, this paper designs OJVis, a visual analytics system for mining and analyzing learning behavior patterns. Comprising five coordinated views - Problem Transition Graph, State Shift Graph, Pattern Cluster Graph, Behavior Evolution Graph, and Code Difference Graph - OJVis enables instructors to comprehensively explore submission logs across multiple dimensions: problem-solving paths, outcome evolution, behavioral patterns, and individual trajectories. The system facilitates precise identification of at-risk students, in-depth analysis of behavioral causes, and data-driven support for personalized instruction. The case analysis based on the real data of the CSUOJ system of a certain school demonstrate OJVis’ significant practical utility and effectiveness in uncovering learning behavior patterns and assisting pedagogical decision-making.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Visual Analytics of Student Behavior Patterns Based on Online Judge Log Data

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Jiaxin Yu,
  • Pengyang Zhu,
  • Guihua Duan,
  • Ping Zhong,
  • Yu Sheng

摘要

As a crucial practical platform for programming education, Online Judge (OJ) systems directly reflect students’ knowledge mastery and comprehension through their problem-solving performance. Existing evaluations primarily focus on pass/fail status and ranking statistics, while overlooking the learning insights embedded in behavioral data such as submission timing, attempt frequency, and keystroke behaviors. To address this gap, this paper designs OJVis, a visual analytics system for mining and analyzing learning behavior patterns. Comprising five coordinated views - Problem Transition Graph, State Shift Graph, Pattern Cluster Graph, Behavior Evolution Graph, and Code Difference Graph - OJVis enables instructors to comprehensively explore submission logs across multiple dimensions: problem-solving paths, outcome evolution, behavioral patterns, and individual trajectories. The system facilitates precise identification of at-risk students, in-depth analysis of behavioral causes, and data-driven support for personalized instruction. The case analysis based on the real data of the CSUOJ system of a certain school demonstrate OJVis’ significant practical utility and effectiveness in uncovering learning behavior patterns and assisting pedagogical decision-making.