How would knowing spammers email address improve spam detection algorithms?

Question

I am new into Artificial Intelligence field and I am working on the classic example of Spam detection using classification. I am using Naive Bayes algorithm as well as SVM.

While working on them it came to my mind that having the email addresses of who has sent the emails as an extra feature would improve the accuracy of my algoithm as we will have some extra info besides the content of the email itself. Would it be true? And how could I use this extra feature? Should I just add it to my email content?

Answer 1

Please accept my apology as it is not really a data science answer.

A lot of spammers are using spoofed emails address as sender, so that field alone might not be very helpful as a dimension. The number of possible values also could make it impractical to be used (i.e., 1 hot-encoding) as a feature unless you preprocess it somehow. Sender domain, sender server reputation scores, etc. would probably be more helpful than the email address by itself.

How spammers spoof your email