The Application of Incomplete Information System of Rough Set Theory in Spam Filtering

The spread of spam brings a great deal of disturbance to e-mail users. In order to improve the efficiency of spam filtering and solve the problem that some attributes extracted from the e-mail header are null, we first applied the knowledge reduction algorithm and the sample recognition algorithm based on incomplete information system of Rough Set theory to spam filtering on e-mail header. In this paper, we proposed an improved non-symmetric similarity relation and defined the attribute complete importance for attribute reduction. Experimental results showed that the algorithm got higher recall and precision, lower fake recognition than other algorithms.