Discretized nested-bigram features distinguish GPT-rewritten Java code from human code with 96-99% accuracy on the authors' new datasets.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Detection of LLM-Generated Java Code Using Discretized Nested Bigrams
Discretized nested-bigram features distinguish GPT-rewritten Java code from human code with 96-99% accuracy on the authors' new datasets.