Plagiarism Detection Based on Complex Feature Set

Aiming at the academic paper plagiarism detection,it is proposed to adopt the complex feature set(CFS) on document representation,and extend the unification operation of CFS. For different features on different levels of document hierarchy structure,it is proposed to use deny functions and probability functions to detect plagiarism.Unifying the shingling method,to take account of different part of document has different importance,defined 3 probability functions based on paragraph weight,and formed a more perfect document plagiarism detection system.Experiments show that the method has produced good results.