Background
I researched a detection of spam sent out using bots. After studying different techniques used by bots to deliver spam, I have no generalised solution to detect spamming at the source network (where bots are residing) in different scenarios (Ex:Using Mx record to directly spam the destined mail-server, using open relays, using webmail services, using open proxy, using spoofed IP address). So, I moved to a receiver level, where all mails are delivered because at receiver side we simply have to say if a mail was a spam or legitimate one, according to some heuristics.
Query
However, could anyone suggest some generalized approach to detect spamming at source network and if possible provide some remarks which would be good, detecting spam mails at the receiver side or the spamming at the source network, where all bots reside?