JdeP said:
Whilst following the Kaspersky Internet Security 2010 SPAM trining sequence I
accidently selected the wrong folder groups in Outlook 2007. Instead of
selecting the spam containing folders to train the software I inadvertently
selected the non spam folders. With a result that KIS 2010 went through the
entire Inbox of ten years content prefacing every single Subject Heading of
every single mail with the designation [!! SPAM].
It's not even my mailbox, it's a client's.
I have a query up on the Kaspersky Support Forum... does anyone know of a
tool (apart from ME that is) that will allow the re re-naming of all the
Subject Headings by removing the [!! SPAM] preface.
This "client" has both 10 years worth of ham *AND* 10 years worth of spam?
Not likely. More likely is that this client has 10 years of ham and very
little saved spam. So you training the Bayesian database (or whatever
anti-spam scheme is employed by KIS) with a high count of ham and very
little spam will severely skew the weighting of keywords in that database.
Also, really old spam is worthless in detecting NEW spam. Any decent
anti-spam program will have an expiry to remove antiquated weightings from
the database, and 10 years is far too long. Your import of ham should not
extend further back than how old is the spam you intend to import. Balance.
There is also something very wrong in organization in Outlook if the client
has been saving 10 year's worth of e-mails in just their Inbox folder. This
is the most changing folder and saving that many items in it can end up with
corruption of the .pst file. E-mails that are to be held [semi-]permanently
should get moved into user-defined holding folders.
If KIS includes DNSBLs (DNS blacklists of known spam sources, like SpamHaus
or SpamCop) then use that scheme to detect spam. Far more accurate than the
guessing scheme of Bayesian keyword weighting. If KIS employs both DNSBLs
and Bayes then order the schemes to first use DNSBLs, any rules you define,
and move Bayes to the end of the detection scheme list.
So where did you save the backup copy of the client's .pst file before you
started operating on it? Obviously this client won't see you as the pro
that you claim to be if you don't backup their data before putzing around
with their host's configuration. No backup means you aren't the pro you
imply by claiming the user is a "client". If you didn't save backups of the
..pst file (or, better yet, all files) then ask the professed "client" if
they have a backup. If you don't backup then you deem your (or their) data
as worthless or reproducible. I partly hope that you don't find the utility
that you are trying to have us find for you (more laziness on your part).
Rather than using us to cover your butt for a bonehead move by modifying
live data, the client gets to know they aren't getting the professional
service for which they are paying.
Okay, after berating you for fucking up the client's data, one suggestion is
that you write a VBA macro to use in Outlook that iterates through each
MailItem to modify its Subject by trimming out the "!! SPAM" substring.
That would be a topic better suited for a VBA programming newsgroup. There
might also be something usable or adaptable over at
www.outlookcode.com, and
they have their own forums where you could ask for help or perhaps con
someone into writing the macro for you. To save face, you might have to pay
a programmer to cover for your boner.
