CONPLAG

CONPLAG是由Sirius.Courses创建的第一个专门针对编程竞赛抄袭的源代码数据集，包含251对抄袭解决方案和660对非抄袭解决方案，均来自Java语言的竞赛题目。数据集通过精选CodeForces平台的21个问题，收集了4695个Java提交，经过多轮手动标记和工具过滤，形成了包含911对解决方案的数据集。CONPLAG旨在为评估和改进源代码抄袭检测工具提供基准，特别关注竞赛环境中的模板代码问题。CONPLAG is the first source code dataset specifically targeting programming contest plagiarism, created by Sirius.Courses. It contains 251 plagiarized solution pairs and 660 non-plagiarized solution pairs, all sourced from Java-based programming contest problems. The dataset is constructed by selecting 21 problems from the CodeForces platform, collecting 4695 Java submissions, and forming a dataset of 911 solution pairs through multiple rounds of manual annotation and tool-based filtering. CONPLAG aims to provide a benchmark for evaluating and improving source code plagiarism detection tools, with a particular focus on the template code issue in programming contest environments.

搜集汇总

背景与挑战

背景概述

CONPLAG是首个专注于编程竞赛抄袭的Java源代码数据集，包含来自CodeForces平台的251对抄袭和660对非抄袭解决方案，总计911对数据。该数据集通过多轮手动标记和工具过滤构建，旨在为源代码抄袭检测工具提供基准，特别针对竞赛环境中的模板代码问题。