Abstract: For low detection accuracy and inability to detect plagiarism at the semantic level in the traditional abstract syntax tree based code plagiarism detection methods,an improved code plagiarism detection approach based on abstract syntax tree is designed.This approach achieves higher plagiarism detection rate through TF-IDF weighted simplified syntax tree.Moreover,feature extraction and similarity calculation methods based on weighted simplified abstract syntax tree are designed to realize partial detection of semantic plagiarism.Experimental results show that the proposed approach can effectively detect semantic level plagiarism and achieves higher accuracy than traditional tree-based methods.

Key words: abstract syntax tree, code plagiarism detection, feature extraction