T-SQL (SS2K8) :: Ranking Duplicate Contacts Using Multiple Columns

I'm in the process of trying to identify duplicate contacts. I doing this for millions of contacts and have gotten stuck and could use some elegant solutions!

The business rule is this:

Any contact that has the same name, phone and email address are the same contact
Any contact that has the same name, and email address are the same contact
Any contact that has the same name, email address, but different phone are a different contact.
Any contact that has the same name, email address, and a blank phone can be the same contact as one that has the same name, email address, and has an email address
Rank by the DataSource_fk. 1 being the highest

Put another way:

If 3 contacts have the same name, 2 have phone '1112223344' and all three have the email address 'johndoe@gmail.com' they are the same contact and the lowest DataSource_fk should be ranked the highest.

I've used the Row_number over (Partition by) in the past, but am unsure how to deal with the blanks in email and phone.

DROP TABLE [dbo].[TestBusinessContact];
GO
CREATE TABLE [dbo].[TestBusinessContact]
(
[TestBusinessContact_pk] INT IDENTITY(1,1)NOT NULL,
[Business_fk]INT NOT NULL CONSTRAINT DF_TestBusinessContact_Business_fk DEFAULT(0),

[Code] ......