HybridCodeAuthorship is a new benchmark dataset of interleaved human-AI Python code that shows existing detection algorithms reach at most 0.48 F1 at chunk level and 0.56 F1 at line level.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
HybridCodeAuthorship: A Benchmark Dataset for Line-Level Code Authorship Detection
HybridCodeAuthorship is a new benchmark dataset of interleaved human-AI Python code that shows existing detection algorithms reach at most 0.48 F1 at chunk level and 0.56 F1 at line level.