Monday, November 16, 2009

Ancestry.com's Name Authority Dictionary

During the SDGS seminar on Saturday, Suzanne Russo Adams mentioned the Name Authority Dictionary that Ancestry.com has recently developed in order to help searchers find their targets with given names and surnames that can be spelled several ways.

Her examples were:

* Elizabeth - there are 900 variations
* Catherine - there are 862 variations
* William - there are 384 variations
* Benjamin - there are 355 variations

* Myers - 238 variations
* McCoy - 219 variations
* Bailey - 127 variations

The "New Search" engine includes the name variations when it performs a "Ranked Matches" search. I don't know if "Old Search" uses it or not. I doubt that "Exact Matches" uses it in either Old or New Search.

I went looking on the Ancestry.com website for more information about it, and didn't find a lot. In the popup box for Getting the most out of new search, there is this paragraph:

"Ancestry automatically looks for common nicknames, abbreviations and other alternate spellings for you. For example, a search for "Bill Smith" might return "William Smith", "Wm Smith", "Bill Smyth" or "B. Smith". An exact name match is the closest match, and therefore the most relevant, followed by common misspellings, nicknames, and other variations. It’s important to remember that often times names are misspelled or mis-transcribed on the original records or in our indexes, so looking for alternate spellings can sometimes help you find a good match, even though the name may look wrong."

It's not clear if this applies to "Old Search," or only to "New Search" with "Ranked Matches." I guess I can try it out later and post about it before Suzanne or Anne or Tony or Laura reads this and provides me the real scoop on the capability! Where's my to-do list?

Of all the search "improvements" that Ancestry.com has rolled out over recent years, this is one that seems to be very useful. The user has the ability to click on "Exact Matches" for any search field in the "Advanced Search" mode. I think that a user should keep the "First Name" field unchecked so that the Name Authority Dictionary can work its wonders. It's probably helpful for the "Last Name" field also, although the Soundex capability is used there also in the "Ranked Matches" results.

1 comment:

Geolover said...

'Old Search' evidently does not incorporate the name-variation-cluster approach. When I search for an 'Elizabeth' it does not retrieve 'Elisabeth', although the wildcard can be used to search for 'Eli*beth' to include also the occasions where it is rendered 'Elisibeth'.

The wild cards tend to disrupt New Search's workings. Maybe that is part of the reason for the Name Dictionary.

Now if they could only fix place names, and get the search engines to stop reversing surname-firstname in results. E.g., when searching for Stanley Morgan, I have no interest at all in Morgan Stanley.