Chrome Autofill often create duplicate profiles. These profiles differ lexically, but have identical meaning to the user. While we do perform some level of deduplication, it is insufficient.
We can improve Chrome's ability to detect and resolve duplicate autofill profiles by adopting the following approaches:
- Normalize fields before comparison to fold case, remove diacritics,
remove punctuation and collapse or remove whitespace.
- Smarter name comparison to account for compound names and the use
of initials. For example, John Quincy Public and John Q. Public.
- Broaden phone number matching to include missing extensions or area
codes.
- Use bag-of-words comparison to compare addresses. This accounts for
different orderings of the information. For example, 23-1 Main St vs
23 Main St, Apt 1
- Use region specific string replacement rules to unify variations of
the way an address field may be populated. For example, Street vs
St, Avenue vs Ave vs Av, First vs 1st, etc.
Comment 1 by rogerm@chromium.org
, Jun 10 2016