Recognition on source codes similarity with weighted attributes eigenvector

The paper puts forward a new method to identify similar C codes based on weighted attributes eigenvector. According to the characteristics of physical and structure attributes of C codes, the weighing theory of attributes eigenvector is adopted to strengthen the infection of attribute elements with covert codes, which are based on the theory and method about existing kinds of attribute eigenvector in C codes. This can also improve the quality and veracity of the similarity recognition. Experiments show that the method can identify the integrated similarity with the three aspects of space characters, parameter variables and sentence sequences. Further more, comparing with ideal calculating results, the measure precision of integrated evaluation can reach to at least 94% in the cases of middle and high similarity.