It would be very useful if a formula was established to determine the likelihood that two users are duplicates within the merge duplicates section. For example: first name, last name, and dob fields all match so the likelihood that these two users are duplicates is 95% vs. the only field that matches between another set of users is the email address so their rating is only 10%.

I envision particular fields/combinations being weighted more than those that are commonly duplicated (email, mailing address, etc.), along with the ability to filter and sort through the full list of potential duplicates. This would add significant functionality to where the utility stands right now.