funny question. as there is no "this is spam" definition, there can't be a clear answer to your question. there are hundreds of different ways of spamming. if you just need one simple example, it's hidden text.
Actually, there is, and the basic methods are quite similar (checking for textual analysis, hosting, domain, content, etc).
The latest antispam tools are called bayesian semantic filters - basically, it's an attempt to program "natural" language traits into computers, allowing them to detect machine generated text, dup content, etc.
Believe me, these spam filters are a little more sophisticated than most people believe :)