Trigonometric words ranking model for spam message classification

The significant increase in the volume of fake (spam) messages has led to an urgent need develop and implement a robust anti-spam method. Several current systems depend mainly on word order message determining spam message, which results system's inability predict correct type when changes. In this paper, new framework is proposed for filtering that does not word's position called Trigonometric Words Ranking Model (TWRM). TWRM based restricting spammers over network by measuring theta angle, relationship between weight spam. classifies calculating rank each places corresponding class. words derived from their frequency entire data category. method applied three datasets messages: UCI email, Enron spam, TREC data. model proven as more efficient than Minhash vector space models. Moreover, performance provided better retrieval time defence, reflected accuracy (99.64%), higher (88.79%) (92.59%).