Tuesday, February 10, 2009

Testing different genealogy search engines - my perspective

I posted What do you want in a search engine? three weeks ago, which listed some of my "wish list" and solicited comments from readers. Several of you responded with suggestions - Thank You! There are many more comments about Searches on the Ancestry.com blog and on the Ancestry.com message boards.

My plan three weeks ago was to conduct tests on a number of web sites to determine the scope of the Search capabilities on these web sites. A good thought, I guess, but I struggled to figure out what criteria to use, and what to test with, and still have not come up with a good answer.

Any comparison of genealogy database search engines is difficult because the databases on different web sites does not have, in general, similar content. I could do a comparison of, say, the 1860 U.S. census, on Ancestry.com, Footnote.com, FamilySearch Record Search and HeritageQuestOnline and have a decent comparison. I will probably start out with that. That comparison would not consider the other databases on those web sites - the military, newspaper, vital records, etc. And that comparison would not consider other web sites with database search functions, such as WorldVitalRecords, GenealogyBank, GenealogyToday, etc. However, other web sites could be tested using another database that demonstrates the versatility and capability of the Search function.

The Search comparisons will include:

* Exact search vs. Fuzzy search ("Fuzzy" meaning variations in names, dates and locations that might result in a match even if the record was enumerated or indexed poorly).
* Use of a specific database vs. many or all available databases
* Use of Wild cards in names
* Use of dates and locations as search criteria

In this exercise, I need to be careful to assess only the Search function, and not the Index function. The Search function can work only on the data that it is provided - the GIGO principle (Garbage In, Garbage Out). Errors by enumerators, transcribers and indexers is a problem for the Index function, not the Search function.

What suggestions do you have? I welcome them, and if you have any other suggestions about your Search "wish list," please comment on this post or the earlier one.

No comments: