Wednesday, February 18, 2009

Genealogy Search Engines: - Post 2: Old Search, Fuzzy Matches

I posted my Search evaluation criteria, the questions I want to answer and the targeted search I want to pursue in this series of posts in Evaluating the Search Engines - Evaluation Criteria.

The first post in the series was about's Old Search, Exact Matches. I won't repeat the discussion about the four defined Search criteria - it's in the first post.

This post demonstrates finding the Isaac Seaver (born 1823 in Massachusetts) family in the 1860 US census using's Old Search function with Fuzzy Matches (i.e., not Exact Matches). We start on the Old Search Home page with the information filled in the search boxes (First Name, Surname, Year, Birthplace) and the Exact Matches box unchecked:

Note that this starting Search box looks a little different once the Exact Matches box is unchecked. Lines appear for Birth and Death years and places.

Clicking on the "Search" button (Click 1) provides the list of matches found by the Search engine:

The top match is the 1900 census which has the right birth year - 1823. The next three matches are the 1880, 1870 and 1860 census records with a birth year of 1824. These are all for my Isaac Seaver, so I want the 1860 census record. These first four matches have four Stars (presumably because the four requested search criteria are exactly right or very close).

Before we leave this page, I want to note some of the other "matches" resulting from the "fuzzy" search. The next entries (all three Star, meaning they matched three of the four requested search criteria) are:

* U.S. Civil War Soldier Records and Profiles for Isaac Seaver 3rd born in 1824 - this is the right Isaac Seaver. It had only three Stars because no birthplace was listed.
* the 1880 census record for Isaac Seaver born in 1809
* the 1870 census record for Isaac Seaver born in 1810
* the 1860 census record for Isaac Seaver born in 1808.

The next part of the 50 matches on this page is shown below:

By the time we get down to the bottom of the screen above, we are still getting Isaac Seaver matches, but the birth year and/or birthplace are going further away from the requested 1823. There are 39 entries for Isaac Seaver on this page of 50 matches - all with three Stars (because they don't match the birth year). The last 11, and all of the unknown remaining matches (Ancestry doesn't tell us how many there are) are for people with a different name. All of the 11 on this page have three Stars because they match three of the search criteria, but not the given name.

Back to my specific search: I want to see if that is the right Isaac Seaver for the 1860 census, so I run my mouse over the link, and see the popup box that shows the family:

That's the right family, so I click (Click 2) on the "View Image" link and get the Result Summary page shown below:

Scrolling down a bit, the source citation is shown below the summary of the family:

I want to see the record image, so I click on (Click 3) the "View original image" link and see:

Looking at my four evaluation questions, the answers are:

* Does the Search find the record I'm seeking? YES
* How many clicks does it take to find the record? Three (3)
* Does the record provide a source citation? YES
* Was the process easy to use? YES

Did you note that the Old Search - Fuzzy Matches method used only three clicks to get to the image, while the Old Search - Exact Matches method used four clicks? That was a surprise to me.

Of course, if this had been a John Smith, or another common name, rather than Isaac Seaver, the Fuzzy Match search process may not have been as easy to use. There might have been more clicks to get to the image, but that may have been the case with the Exact Match search also.

As in the Exact Match search with the Old Search interface, the filled-in Search box appears at the bottom of the Search Match page. If I had wanted to change the search parameters at that point, I could have easily.

Next up is to look at the New Search interface with Exact Matches or Fuzzy Matches.


anne said...

Hi Randy.

I've been watching your posts with interest -- of course!

Just one note on the fuzzy vs exact: you don't have to do all fuzzy or all exact. One of my favorite tricks to find who I am looking for is to fill in the most exact value I can in the "Lived In (Residence)" field and click exact.

For example, if you are looking in Chicago, first start with:

"Chicago, Cook, Illinois, USA" and check exact only on the Lived In. That will filter all results to to records that are for Chicago. If that is too narrow, you can then try "Cook County, Illinois, USA" and then "Illinois, USA"

That will help you with all of the pesky ancestors who didn't have unique names.

Keep up the good work -- this is a great set of posts.

Anne Mitchell
Product Manager,

Russ said...


I've given you the KreativBlogger Award - Visit to receive your award.


Familytreeservice said...

well done, great insight into searching on ancestry. one problem I have always had is putting on exact information for say, Mabel Molesworth (this is a serious name, honest), born 1890, Redditch, Worcestershire, England. Fairly specific details and name I think you will agree. However, as her name is recorded as 'Mable' in one census, her result comes after about 5 pages, behind every Molesworth name in the whole of England let alone in Worcestershire. I've noticed this happens for William (Wllm) and Thomas (Thos) a lot. Any tips on avoiding that?