A COMBINED METHOD FOR DETECTING SPAM MACHINES ON A TARGET NETWORK
Tala Tafazzoli, Seyed Hadi Sadjadi · 2009
The HITS and PageRank algorithms and K-Means clustering algorithm are two main methods for detecting spam machines. In PageRank algorithm, it is proposed to calculate weights based on different factors. Correct selection of weights has important role in the accuracy of the algorithm. In this paper, we propose a good method for convenient selection of weights. We first executed the K-Means algorithm on the traffic of a big target network and divided IP addresses to two parties, normal and anomalous, and assigned a weight to the IP addresses of anomalous party which is used in calculating the energy rank of the second method. With executing the second algorithm, we found a larger set of IP addresses of spam machines and found that we have increased the accuracy of the algorithms perceptibly.