[Rspamd-Users] cold startup question

[G.W. Haywood]

Hi there,

On Sun, 30 Jun 2024, Jeff Peng via Users wrote:

> as the cold startup, where to get lots of spam/ham messages for training 
> Bayesian?

The profile of spam messages is different for different sites.

The idea of training the Bayesian filter is to train it on the spam
which arrives at *your* site.  The spam which arrives at a different
site is unlikely to be especially useful to you.

If you want to filter based on the spam received by other sites you
should be looking at a different approach.  The simplest and most
effective is probably to use a few DNSBLs.  As an indication of what
is available, try replacing [[IP_ADDRESS]] with the IP addresses from
a few samples of your spam here:

https://multirbl.valli.org/dnsbl-lookup/[[IP_ADDRESS]].html

Be aware that Valli.org makes no representations about the usefulness
or otherwise of those DNSBLs listed.  You need to try them or research
them yourself.

But don't worry, you will soon get plenty of spam to train the filter.

-- 

73,
Ged.