Pages

Saturday, May 16, 2015

What do the automatic genealogical search programs find?

Four of the largest online genealogical documents websites, FamilySearch.org, Ancestry.com, Findmypast.com and MyHeritage.com, have all implemented automatic record hint capabilities. What do these four programs find for the people in your ancestry? Are they are finding the same records or are the results different? Are the documents they find useful for research into your family lines?

In order to use all four programs, you must be registered with each of the four. FamilySearch.org is free but the other three are paid subscription programs.

Since I have been using all four programs for quite a while now, my impression is that they are all extremely effective in accurately finding documents pertaining to particular ancestors. Overall, their accuracy is, in some cases, astounding. The general limitations are rather obvious. The programs cannot find records that are not in their own particular set of documents or collections. Their accuracy diminishes as they go back in time, just as the number of online, available documents diminishes. They have a difficult time distinguishing between people with very similar names, dates and geographic information, just as human researchers do. All in all, they work very well going back about 200 to 250 years, but then their usefulness decreases rather rapidly.

Each of these four programs requires the user to enter some basic information before the automatic search process can begin. In other words, you have to have a family tree on each of the four programs to take advantage of the automatic search capabilities. This fact creates some rather serious perceived issues, including the need to somehow keep all four of the family trees synchronized and the difficulty of moving information from one tree to another. This problem becomes even more complicated if you elect to maintain your own family tree on your own local genealogical database program. Then, in effect, you have a minimum of five family trees to contend with.

In the past, I have written about the process of determining what documents each of these websites have in their collections. Here is a brief summary of where you should look in each of the four programs. The idea here is to ascertain whether or not the programs have any documents that may help you find your ancestors in a specific location. However, you should always be aware that there may be pertinent documents in collections that you are completely unfamiliar with and would never have searched.

Another point is the utility of doing manual searches on each of the programs. In general, each of the programs has a superior ability to find some types of documents using their automated search functions than your own ability to find the same documents by manually searching. But I have found that there seem to be some documents that can only be found by manually searching in the collections individually.

Here is the summary:

FamilySearch.org
All of the records on this website are listed in the FamilySearch Catalog. Those records that have been digitized and are in the Historical Record Collections are indicated by a link to the collections. However, the only documents that are currently available for Record Hints are those that have been indexed and added to the Record Hints Search. These records that are automatically searchable constitute only a small percentage of the entire set of records on the website. The number of automatically searchable records is constantly increasing as new collections are added to the automatic Record Hints category.

Ancestry.com
All of the records on Ancestry.com are indexed and searchable. There are no obvious limitations on the number of records from the entire collection that is included in the automatic "Shaky Green Leaf" record hint technology. However, you may want to review the Ancestry.com Card Catalog and become acquainted with its contents.

Findmypast.com
The collections available from Findmypast.com are more focused on the United Kingdom and related records than other parts of the world. Of the four programs, their search capabilities are the most rapidly evolving. The records on Findmypast.com are listed in the A-Z of Record Sets.

MyHeritage.com
Each of the three other programs, start you out with a search for records for your ancestors. MyHeritage.com, more than the other three programs, relies almost completely on its automated search capabilities. You can search for records directly, if you wish to do so, and the list of collections or databases in the program is available, but doing your own searching is not nearly as effective as letting the program do the searching. The list of programs is available by geographic region under the "Research" tab on the home page of your family tree.

Choosing an ancestor to use as an example is rather difficult for me. Since I have had all of family tree data in each of the programs for an extended period of time, they have all had ample time to find suggested records. There is no way to compare the number of hints made available by all four programs for my entire record, since FamilySearch.org and Findmypast.com do not provide you with a "total" number of hints they have found. If you were to choose one particular ancestor, the choice itself would likely determine the outcome. For example, if I chose an ancestor from England, then Fimdmypast.com would have the advantage in any comparison, simply by virtue of the fact that they have the most indexed records from the U.K. In each case, the number of records found will be determined by the match between your ancestral lines and the particular types of records in each of the programs.

From another standpoint, the records the do find may be extraordinarily valuable for one ancestor and cumulative for another. Without a method of evaluating the effectiveness of the hinting process, there is also no way of comparing raw numbers. It also seems that no matter how accurate the programs' search capabilities may be, they are still liable for a number of "false positives" or record hints that do not apply to your ancestor.

Finally, record hints are not a substitute for careful evaluation of the record contents and application. Just because the programs are accurate and helpful does not mean they are always right. You have to examine each record carefully before incorporating the information into your file. You must always remember that the records themselves may not be accurate.

2 comments:

  1. Hi James. I just wanted to let you know that I believe there is a limit on the record sets featured in the Ancestry shaky leaves. In their RootsTech 2015 presentation, Crista Cowan and Juliana Szucs stated that the hints are taken only from the top 10% most used record collections. I believe that is what was said though someone with inside info can correct here if I am wrong. I can't find my notes from that presentation but I am sure there is a limitation because I remember being quite shocked. Likewise, in Jen Baldwin's FindmyPast class, she said they were only doing BMD hints for now, though that may have changed since February (that is in my notes I did find). So clearly we aren't always being told what we are and aren't getting so the best practice is to manually search everyone and not rely upon these automatic searches to catch every record, especially since they often don't catch alternate spellings, etc.

    ReplyDelete
    Replies
    1. Thanks for your interesting insight. I am not suggesting that the automatic search features are an alternative to further research.

      Delete