How do you propose to make such relations between the keywords, and how to
mitigate false positives?
Naïve Bayes deals with words independently. If we want to link between words, I
think we are into Natural Language Processing (NLP).
If you have any good thoughts please share.
---
Mahmoud
Hello Marc,
Am 2010-10-03 09:21:02, hacktest Du folgendes herunter:
I think we could catch a lot more spam if we had some God/Christian
rules.
Hey, thats cool!! I need rules for Allah/Islamic stuff!
Problem 1: They are written in arabic, urdu and two other
languages I do not
As I said in my original message. The rule itself would have no
points. It would be combined with other rules like large sums of money
and transfer and I'm someone you don't know.
On 10/3/2010 10:27 AM, m...@khonji.org wrote:
How do you propose to make such relations between the keywords,
On 10/3/2010 12:09 PM, Michelle Konzack wrote:
Hello Marc,
Am 2010-10-03 09:21:02, hacktest Du folgendes herunter:
I think we could catch a lot more spam if we had some God/Christian
rules.
Hey, thats cool!! I need rules for Allah/Islamic stuff!
Problem 1: They are written in arabic,
On Sun, 3 Oct 2010, Marc Perkel wrote:
As I said in my original message. The rule itself would have no points.
It would be combined with other rules like large sums of money and
transfer and I'm someone you don't know.
You mean, like the ADVANCE_FEE and MONEY_FRAUD rules?
--
John Hardin
