Social context based effective filtering of spam messages from e mails and online social networks

Abstract

Now a days Electronic communication is an important medium and an inevitable way for official and personal communication. E-mails and Online social networking sites such as Facebook, Twitter, and LinkedIn have shown an unbelievable widening in the last decade. The spammers utilize the e-mail medium and Online Social Network sites to unroll spam messages due to its fame and also use various procedures to spread spam. The identification of spam must be well fortified enough to detect unsolicited messages and deter spammers. The text-based or collaborative methods are commonly used for spam message detection. So, the spam identification procedures are needed to improve the accuracy and reduce the false positive and false negative rate for identification of spam in e-mails and social networks. The proposed model for e-mail spam classification considers a powerful spam filtering technique which includes both social network and email factors in addition to the email data analysis in e-mails. The incoming emails are subjected to header parsing for finding the trust and reputation of senders with respect to the receivers. The keyword parsing is also applied to find the topic of interest using Latent Dirichlet Allocation with Gibbs Sampling method. Optical Character Recognition method is applied to find the e-mails with image spam. The degree and strength of the connection between the users from the social networks are also considered along with the email factors for message classification. The Logistic Regression is used to combine all the independent input features to improve the accuracy. The experimental results vividly shows that the significant performance of the proposed E-mail and Social Network based model with Bayesian Filter (ESN-BF) achieves an accuracy of 96.5% for spam message classification in emails when compared with the other state-of-the-art methods. The proposed model identifies the spam messages from the online social network users walls. The social context parameters such as trust between the users and the strength of the relationship between the users are considered in the proposed model for spam message classification. The intercommunication factors between the users are used for strength calculation. Spam template is generated based on the majority merge operation on the spam messages during the training time, and a comparison of the spam templates is performed with the incoming messages during the testing time. The updation of the trust value is performed after the message classification. The experimental results demonstrate that the proposed Template and Social Context based SVM Filter (T-SC-SF) with Polynomial Radial Basis kernel provides an accuracy of 99% in spam classification when it compared with the other state-of-the-art methods. newline