Newdora

1.0 Problem Statement
Creat  a mail client with a spam filter operating at the local level that
has a low level of false negatives and false positives. 

-What is Spam?
Spam is a difficult problem to deal with, or even define. The small number
of people who buy penis enlargement pills may not consider such messages
as spam. Attempts to prevent or alleviate spam through legal or technical
measures have met with limited success. Spam is a menace to society,
because the gain to the spammer is large compared to his costs, but small
compared to the costs to society in general, specifically bandwidth use
and a small amount of time from millions of recipients.  
A promising new solution to spam was
recently proposed by Paul Graham
(http://www.paulgraham.com/spam.html).

-Bayesian Filtering
This "Bayesian" algorithm requires input from the user. When they receive
a message they may mark it as spam. Messages are filtered by measuring how
often each word in the message appears in spam versus non-spam
messages. It requires incorporation into a mail client to be convenient
and effective. 

-Functional Requirements 	 
The mail client side should include: 
	-Ability to compose new email, supporting CC, BCC, and multiple
	attachments. 
	-Ability to receive email. 
	-Kerberos support. 
	-Stored messages can sort by sender, date, and subject. 
	-Address book. 
	-Add a signature to messages. 
	-Spell check. 
	-Allows user to mark message as spam. 
	-Messages filtered as spam stored in separate mailbox. 

The spam filter side should include: 
	-Uses Bayesian algorithm to filter spam. 
	-No more then 1% false negatives after first 100 non-spam and 10
	spam messages are received.  
	-No more than .1% false positives after first 100 non-spam and 10 
	spam messages are received.  


-Priorities
	1)Mail client can send and receive email. 
	2)Spam filter fully implemented. 
	3) Kerberos support. 
	4) All others.