Thursday, April 25, 2013

Response from on my Wild Card Search Frustrations

In my post Home Page Changes and Wild Card Search Frustrations on Tuesday, I described my frustration over some wild card searches don't work on and wondered why.  

I received an email yesterday from Matt Deighton at (Specialist, Public Relations)  with a note from Katherine Nester (Director, Product Management) that explains what is happening:

Dear Randy,

We recently read your blog post and wanted to send you a quick note to clarify a few things and thank you for the feedback on the changes to the search functionality on the logged in home page. We have made changes and will be continuing to make changes to both increase the performance of the site and to begin to build a service that works for both the new user to and the seasoned genealogist. As Tim mentioned at RootsTech, making the site work for both groups is a focus for us this year. 

The logged in home page was one of the slowest pages on the site and was still talking to outdated systems, limiting our ability to make improvements. The recent changes, including the search form, improve the performance and enable us to evolve the experience to address our different audiences.  We wanted to get the performance improvements out as quickly as we could and additionally start getting feedback from our users, like yourself. This feedback will help us continue to tailor and evolve the experience so that it works well for both audiences and we very much appreciate your use cases and ideas.

In regards to the problem you have previously reported to us on the wild card searches, we have researched this issue and defined a solution. The reason your search does not work in the Category view is due to the extra server-side complexity of the wildcard query combined with sorting for such a broad result set.  (Note that it does work in the Relevance view.) That’s  a current limitation of our search system which is querying 30,000+ collections with 11 billion records representing terabytes of unique field data. We realize that this type of feature is beneficial to our users, which is why it is on our list of search functionality to improve as we make adjustments to our product experience. 

Here are 2 suggested workarounds:

1. Alter your search such that your wildcard produces a slightly more narrow result: isa* s?v?r   -or-  isa* seav*  -or-   isa* sea*r

2. Switch to the Relevance view before doing your search, then use the “Narrow by Category” functionality to get into a specific category, THEN you can switch back into the Summarized by Category view.

As always, we appreciate your feedback and welcome addition comments you have that can help us better understand your experience with the site.

Best regards,
Katharine Nester
Director Product Management

I really appreciate that Katherine took the time to explain how my problem occurs, and to provide some workarounds for it.  Both workarounds work for me, so I will try to use them as Katherine suggested.

Her answer about why the problem occurs implies that there are different search processes at work for the Relevance view (perhaps newer and faster) and Category view (perhaps relying on an older search algorithm similar to Old Search). 

While trying to do my own analysis of the problem, I found that doing the wild card search for "isa* sea*" works fine in Old Search with or without exact matches selected (exact matches selected results in a Category view match list).  So that's another option for users.

I won't pretend to know how these search algorithms work.  I'm always amazed that any search can return millions of matches in a few seconds to my computer screen.  As an example, I received over 4.5 million matches for a "john smith" search in less than five seconds.  

As I say in my presentations about, "Ancestry has (IMHO) the most sophisticated and complex search process in the genealogy world."

I do really appreciate the search capabilities, and am puzzled when they don't work as I expect them to!

The URL for this post is:

Copyright (c) 2013, Randall J. Seaver


Jay at 1FamilyTree said...

Randy, I get really frustrated when I hear (or read) comments like "The logged in home page was one of the slowest pages on the site and was still talking to outdated systems, limiting our ability to make improvements. " SORRY, but companies that cater to the small percentage of persons who are on older systems are not only hampering their own growth, but the growth of all their other users!!!

Anonymous said...

Randy, you're much too easy on them. They're feeding you a pile of BS.