2

My master thesis is about "Managing Spam Under IPv6". I would like to use machine learning algorithm on information retrieved from mail headers in order to sort out spams from legitimate mails. To do my testings I obviously need a huge dataset of spams and legitimate emails (hams). Unfortunately I have been unable to find a large enough dataset of IPv6 spams and legitimate emails (hams) on internet datasets. Does anyone know of such a dataset in the public domain?

Alternatively, any ideas in order to create my own dataset would be greatly appreciated! Are there any researchers who work in this domain that I could contact?

For now I am basically using gmail spams from some friends of me, I can find few ipv6 spams but the amount I need is much larger if I want exploitable results from the machine learning ...

PS: I know this is not the best kind of question to ask on steckexchange, since it leads more to an open discussion, but I really need help and this is for the sake of science so please be indulgent ^^

  • Hi @Vrael and welcome to security.stackexchange! I took a bunch of text our of your question to make it shorter and easier to read. If you don't like my edit, you can roll it back by going here. – Mike Ounsworth Apr 6 '17 at 14:05
  • What's unique about IPv6 as a delivery mechanism for spam? Aren't all the interesting bits on the application layer? – schroeder Apr 6 '17 at 14:07
  • 1
    @schroeder IP address reputation management becomes trickier since IPv6 addresses are cheap and you can't easily tell how big a subnet is given to each user. – CodesInChaos Apr 6 '17 at 14:10
  • @CodesInChaos sure, but how then do you run ML on such a vast data range to come up with a reputation model? You aren't going to get anything interesting by inspecting network addresses. – schroeder Apr 6 '17 at 14:13
  • 1
    @schroeder Check this link usenix.org/legacy/event/sec09/tech/full_papers/hao.pdf In this paper they use simple features like AS origin, message length, time of sending, .... I d like to see if it still good under ipv6, that is what I want to use for ML – Vrael Apr 6 '17 at 17:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.