One of the approaches used to identify spam goes like this; if I see a spam message at 8:30 in the morning, I send a checksum of that message to an online database of spam. When you get that message a little later on in the morning, your mail system asks that online database, "Has anyone reported this as spam?". The online database can report back "yes", allowing your mail system to raise the spam score for that message.
See UsingRazor, UsingPyzor, and UsingDcc for details on those supported natively in SpamAssassin.
See iXhash and CustomPlugins for third-party add-ons for SpamAssassin.