Sequential Pattern Mining for Chinese E-mail Authorship Identification

With the rapid growth in computer technology and popularization of Internet, e-mail has become one economical and convenient form of communication. But different types of crime and civil action involving e-mail documents appear which do harm to people´s life and social´s stabilization. So the criminal e-mail´s authorship has to be identified automatically for the purpose of computer forensic. To solve the problem, the appropriate feature extraction and selection methods are essential. Unlike English and other IndoEuropean languages, Chinese text does not have a natural delimiter between words. Word segmentation is a major problem in Chinese text processing. So in this paper, sequential pattern feature mining methods were described without word segmentation. The support vector machine algorithm was adopted as classification algorithm. The experiments on limited samples gained satisfying results, which proved that the sequential pattern feature mining methods were effective.