>  I like this idea. I would suggest to combine implementations for several
>  spam detection mechanisms under a common hood. I imagine a component
>  where you can configure a pipeline of spam detection mechanisms you'd
>  like to use and then make it check a certain text.

I really like the idea of the pipeline! And maybe we can have a set of
pre-configured pipelines for common tasks, e.g. "blog comment
pipeline", where we would try to find the optimal pipeline to detect
blog spam.

>
>  Ideas for spam detection mechanisms I have in mind are:
>
>  - DNSBL
>  - SURBL
>  - Akismet
>  - Bayes
>  - Rule based

This is exactly what I had in mind when I was chatting with Thomas
yesterday. I guess we need to grow the list and make it more
comprehensive. Would something that interfaces into used email spam
detectors be of use(e.g. spam assassin)? or is that too much hassle
and out of the scope to be included in such a component.

- Ammar
-- 
Components mailing list
[email protected]
http://lists.ez.no/mailman/listinfo/components

Reply via email to