enabling ocr capabilities of Spambayes on Windows

I am using Spambayes 1.1a3.
I am using Outlook 2003 in an Exchange 2003 environment.
I am using the Outlook plugin of Spambayes.
I have configured Outlook to show the spam score.
I have downloaded the compiled ocrad.exe and copied it to
the
Windows-directory.
I have downloaded and installed Python 2.5.
I have downloaded and installed PIL for Python 2.5
I have created a file
"default_bayes_customize.ini" in the Spambayes
data
directory (c:documents and
settingsamedee.van.gasseapplication
dataspambayes).
I have put the following lines in
default_bayes_customize.ini:

[Tokenizer]
x-crack_images: True
x-fancy_url_recognition: True
x-image_size: True
x-lookup_ip: False
x-pick_apart_urls: True
x-reduce_habeas_headers: False
x-search_for_habeas_headers: False
x-short_runs: False

[URLRetriever]
x-cache_directory: url-cache
x-cache_expiry_days: 7
x-only_slurp_base: True
x-slurp_urls: True
x-web_prefix:

I have restarted Outlook.
I have _not_ deleted my training database.
I have rescored all messages in a certain folder.
I have looked at the spam clues of several messages.
I have seen no indication that Spambayes is working
differently now, at
least I see no sign of image reading.


What else must I do to enable ocr in Spambayes?

-- 
Amedee Van Gasse
amedeeamedee.be

_______________________________________________
SpamBayespython.org
htt
p://mail.python.org/mailman/listinfo/spambayes
Check the FAQ before asking: http://spambayes.sf.
net/faq.html

    Amedee> I have restarted Outlook.
    Amedee> I have _not_ deleted my training database.
    Amedee> I have rescored all messages in a certain
folder.
    Amedee> I have looked at the spam clues of several
messages.
    Amedee> I have seen no indication that Spambayes is
working differently now, at
    Amedee> least I see no sign of image reading.

    Amedee> What else must I do to enable ocr in
Spambayes?

Depending on the nature of the messages containing image
spam, enabling ocr
may have had no effect.  The obfuscation applied to image
spam now (splotchy
backgrounds, multi-colored text, etc) frequently makes the
text invisible to
the ocr stuff in 1.1a3.  I've worked on some enhancements
(not yet checked
in) and Mark Hammond has some changes to allow gocr as an
alternative to
ocrad as the ocr engine.  Both seem to help, but we've not
done a release
yet.

Skip

_______________________________________________
SpamBayespython.org
htt
p://mail.python.org/mailman/listinfo/spambayes
Check the FAQ before asking: http://spambayes.sf.
net/faq.html

> I am using Spambayes 1.1a3.
> I am using Outlook 2003 in an Exchange 2003
environment.
> I am using the Outlook plugin of Spambayes.
...
> What else must I do to enable ocr in Spambayes?

Work on OCR in the outlook plugin is still ongoing, and no
binary releases
have been made with it enabled.  You must either run from
source-code, or
wait for a new release (and unfortunately I've no idea when
that will be)

Mark

_______________________________________________
SpamBayespython.org
htt
p://mail.python.org/mailman/listinfo/spambayes
Check the FAQ before asking: http://spambayes.sf.
net/faq.html