Explaination of settings

Validate the IP address of commenters.

As an additional method of spam detection, ASB can retrieve available network information of the sender (eg hostname) and evaluate them anonymously. Analysis will be executed directly in your installation without sending data to any external services.

Use regular expressions.

Some spam comments match a certain pattern. ASB can use those to filter spam. Comment text, e-mail address, and IP will be tested against regular expressions predefined by the plugin, or the user. Comments matching a given pattern will be filtered out.

Use a public anti spam database.

This functionality is based on the progressive database of Stop Forum Spam, a publicly available database of IP addresses of active spammers. If a comment spammer is represented in the data pool of Stop Forum Spam, ASB will mark their comment as spam.

Disclaimer: This option shouldn’t be activated on sites within the EU as complete IP addresses will be sent unencrypted. However, that’s somewhat inevitable as spammer databases themselves work with full, unencrypted IP addresses.

All answers above have been translated into English from the documentation in German, errors and omissions excepted. I have no deeper technological understanding than any average user.

The other question i had was, should i still use Akismet?

There’s no need to, so you shouldn’t. Either use Akismet, or use ASB.

Disclaimer: This option shouldn’t be activated on sites within the EU as complete IP addresses will be sent unencrypted. However, that’s somewhat inevitable as spammer databases themselves work with full, unencrypted IP addresses.

I just tried something manually: ASB could (maybe optionally) connect to https://www.stopforumspam.com/api?ip=91.186.18.61 (an example from the SFS page) instead of http://www.stopforumspam.com/api?ip=91.186.18.61.

At least that would take care of the ‘unencrypted IP adresses’ part.

Update: On the regular expression part, I recommend a function hook as described in this gist: https://gist.github.com/sergejmueller/4242142. I’ve been using this sucessfully for over a year now (specific to the kind of spam my wife’s blog gets):

function antispam_bee_add_custom_patterns($patterns) {
	// Pattern for phony author names.
	// Fun Fact: The last one is 'Prada' in Japanese.
	$patterns[] = array(
		'author' => 'moncler|north face|vuitton|handbag|burberry|outlet|dress|maillot|oakley|ralph lauren|ray ban|iphone|プラダ'
	);

	// Pattern for phony web pages.
	$patterns[] = array(
		'host' => '^(www\.)?fkbook\.co\.uk$|^(www\.)?nsru\.net$|^(www\.)?goo\.gl$|^(www\.)?bit\.ly$'
	);

	// Pattern for text containing strings like 'targetted visitors'.
	$patterns[] = array(
		'body' => 'target[t]?ed (visitors|traffic)'
	);

	return $patterns;
}

@simon Thank you!