Monday, August 2, 2010

Searching Collections on FamilySearch Beta and Record Search - they're different!

I've been working quite a bit in the FamilySearch Record Search Pilot over the past few months, and then have been epxloring the Family Search Beta site recently.

Some things puzzle me a lot. For instance, here is the same search done in the same database on the two sites:

1) On the FamilySearch Beta site, the list of "All Collections" looks like this:

There are 289 record collections listed on the FamilySearch Beta site.

I chose the California San Francisco Area Funeral Home Records, 1835-1931 collection to search from the list above. The site noted that there were 43,407 records in that collection.

I clicked on the database, and put "Smith" in the Last Name field (of the Advanced Search box):

The result was 412 matches:

There appear to be no way to filter the matches by gender, first name alphabetized, etc. A searcher could put a first name in the search field on the left sidebar and search again.

2) On the FamilySearch Record Search Pilot screen, there is a list of 430 record collections as of today, with the latest updates or new databases denoted by a red star (which I like!):

I scrolled down to the California San Francisco Area Funeral Homes, 1835-1931 database and clicked on it. I selected the "Advanced Search" box and entered "Smith" in the Last Name field and selected "Exact and close match" in the field next to the "Search" button:

Here are the Search results:

For this search, there were 409 matches. I could refine my search using one of the filters provided (Place, Date, Gender, Last Names, First Names, Role and Event Type), or I could Refine my Search by clicking on the "Refine your search" link.

I wondered why the FamilySearch Beta search had 412 matches, while the Record Search Pilot had only 409. I wondered if there were more if I chose "Exact, close and partial matches" in the RecordSearch Pilot - and I got the same 409 matches. Are there a different number of records in the two collections - that really makes no sense to me. While the difference is small, it does approach 1%, about the same as an indexing error rate in many databases.

What about the order of the matches? You can see by comparing the first list of matches for each search that they list different persons. The organization of the lists is not obvious - they don't seem to be alphabetical on first name, or last name, first name, or on birth year, or on death year. The matches seem to be provided in random order on both sites, but in a different random order! Apparently, there are two different search algorithms.

To recap, the FamilySearch Beta record collection, and the FamilySearch Record Search Pilot record collection have these differences, at least today:

* FamilySearch Beta has 289 collections, while the Record Search Pilot collection has 430. It's not clear if all of the 289 are in the 430 collections in the Record Search Pilot collection, or if FamilySearch Beta has some unique collections not in Record Search Pilot. If I were real energetic, I would compare them, but I'll leave it to others.

* The Search algorithms for the two sites seem to be different, They come up with a different number of matches for a given surname, at least in this selected database (I don't have the patience or time to check several others).

* The order of the presented search matches are also different, and have no apparent logical order.

* The FamilySearch Beta Advanced Search does not have any filters available, while the Record Search Pilot has filters (and they are pretty useful).

* The user can choose "Exact," "Exact and close" or "Exact, close and partial" matches in RecordSearch Pilot, while it is unknown what the search parameters are for FamilySearch Beta.

The gossip on the Internet is that a big change is coming soon to FamilySearch's main website. The speculation is that it will look like the FamilySearch Beta site. Or not. Who knows? Only the FamilySearch Labs folks, I guess.

Whatever it looks like, I sure hope that the site provides the Record Search Pilot filters and search criteria.

Can you tell that I'm confused by all of this? Why are there differences in the search algorithm, the number of matches, etc.?


James Tanner said...

I agree entirely. The differences between the site and the each of the individual sites,,, and Record Search are pretty confusing.

Katharine Ott said...

I also agree. The labs site which you displayed is the one I always go to first as it is much more user friendly with the filters and "hover" descriptions of documents. Most of my Ohio Deaths, 1908-1954, are now directed to the new beta site which I valiantly try to wade through. Since it's beta, there's always hope that they will make some adjustments.

Anonymous said...

The Beta search reminds me a lot of New Search at Ancestry - which I also dislike. I tend to go the pilot site first, and only go to beta for the matching image or for completeness.

Sassy Jane Genealogy said...

Another in agreement here. And let's also throw FamilyLabs into the mix. Ironically, I sometimes get better results at there even though it's billed as "not ready for prime time."

On the plus side, I sometimes get census results that I've never seen at Ancestry.

Sharon said...

I'm totally confused by the sites. Why are there two?
The search on is not very good but the search on is useless.

Sassy Jane Genealogy said...

Sharon, I've had good luck with the Germany Births and Baptisms and the Germany Marriages in the familysearch beta site.

But for U.S. searches, the one branded as Family Labs (
is better - at least it is for my Cook County searches.

OurFamily said...

Each site does use different search technology. This is why the hits are slightly different in nature and order. We released a new version of this technology yesterday which greatly improves accuracy of the results.

In the near future the beta site will have all the collections that are in RecordSearch pilot and more.

In coming months we will be extending the filtering capabilities that you see on the left side bar of the results page to include many of the most used filters in RecordSearch.

One of the goals for the beta site is to unite all the various tools (like those that James mentions) and the data sources under one site and provide a more seamless user experience as one moves from one tool to another.

I encourage everyone to leave detailed feedback on the site about what they like and what is still missing. FamilySearch beta is like a construction project and we greatly appreciate user input as we build it out.