27.06.2013 Views

6th European Conference - Academic Conferences

6th European Conference - Academic Conferences

6th European Conference - Academic Conferences

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Manoj Cherukuri and Srinivas Mukkamala<br />

Figure 7: Overview of the process for the construction of the dataset<br />

Table 1: Top five countries by the number of malicious websites hosted in our dataset<br />

5. Link analysis<br />

Country Number of malicious websites<br />

United States 14790<br />

Philippines 1086<br />

Canada 432<br />

Germany 183<br />

United Kingdom 143<br />

5.1 Outdegree and indegree of malicious websites<br />

The indegree and the outdegree of the malicious websites within the dataset were computed and<br />

plotted two graphs representing the count of the malicious domains versus the indegree and the<br />

outdegree. For computing the indegree and the outdegree of the websites, we considered only the<br />

links among different domains as most of the links within the same domain were identified to be<br />

navigational links. The count versus the indegree and the outdegree graphs are shown in Figure 8.<br />

The outdegree and the indegree of the malicious websites did not satisfy the power law in contrast to<br />

the World Wide Web graph (Watts and Strogatz, 1998).<br />

In an attempt to identify an equation that suites the indegrees and oudegrees of malicious websites,<br />

we identified that the malicious websites satisfy the power law with an exponential cutoff. The Lambda<br />

and the Gamma values of the power law with exponential cutoff equation for the indegree and the<br />

outdegree of malicious websites were identified to be 12.32, 0.9 and 8.32, 1.02 respectively.<br />

Correlation coefficient was measured to verify the fit of these equations. The correlation coefficient<br />

was 0.98 and 0.99 for the indegree and the outdegree respectively signifying a good fit.<br />

<br />

.<br />

<br />

<br />

Where <br />

<br />

is the exponential cutoff and is the power law term (Clustering Coefficient, 2010)<br />

58

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!