关键词: 垃圾短信, 交互单元, 有效交互周期, 入出比, 查准率, 查全率

Abstract: In a time flooded with massive spam messages, clearing them waste a huge amount of effort and time. The mining sent feature of spam messages is the key to solving this problem. On the basis of analyzing current text-message filtering mechanisms, an effective interaction period is proposed by combining the discrete interaction units of the message sender into a consecutive interaction unit according to the essence of median filter. Utilizing the ratio of input to output and Effective Interaction Period (EIP), a general filtering algorism of spam message is built. Experimenting on 20 millions real messages, the recall ratio of the proposed algorithm is 99.51% and the precision ratio is 49.90%. The experimental results indicate that the novel algorism greatly enhances the efficiency and velocity of detection, which can be applied to spam messages real-time intercepted technology.

Key words: spam message, interaction unit, EIP, the ratio of input to output, precision ratio, recall ratio