Friday, January 23, 2009

What do you want in a search engine?

Every researcher complains about them, but I haven't seen many lists of what people want in a genealogy database search engine.

I've been using Ancestry.com, HeritageQuestOnline, WorldVitalRecords, Footnote, FamilySearch, MyHeritage, WeRelate and many other web sites - some for months and some for years. I've gotten to "know" their search engines intimately. I would like each of them to improve their capability.

Which web site has the absolute "best" search engine, in your opinion? And why do you think it is the best?

What capabilities do you want to see in a search engine?

Here are some of my search capabilities that might be considered:

* Exact search on as many fields as possible
* Fuzzy search on fields, with ranges for dates and ages
* Fuzzy searches should include alternative spellings for names (not necessarily soundex or metaphone).
* Ability to choose between exact and fuzzy search on each field

* Ability to use wild card symbols for names and localities
* Ability to use the ? symbol for one letter in a name
* Ability to use father's first name and birthplace, mother's first name and birthplace
* Ability to use spouse's name and birthplace
* Ability to use sibling's name and birthplace

* Define proximity to a name in a keyword (i.e., page, paragraph, line, next to, etc)

* Ability to search a specific database, not the entire collection

* Put editable current search box on current results page to minimize clicks and page changes.

Those are my "wants" for a search engine just from my mental checklist. What do you want? Tell me! I'll create a master list. If you want to blog your own list, please do so, but please tell me where you've written it.

Frankly, when I'm searching I want to minimize my clicks and have maximum flexibility to choose search terms and ranges. The quicker I can find a result the happier I am.

Please note that this list should not consider what you want in an index - only what you want in a search engine. The index problem is a whole different mountain to climb!

In the next several weeks, I am going to write a series of posts that demonstrate what each site uses from the so-called "master list" so that my readers understand the differences and can use each database effectively.

6 comments:

GeneaDiva said...

I especially wish Genealogybank would allow one to search just one specific newspaper or other database. I get way too many hits when I select by state for historic newspapers and also the Historical Documents. Many times I just want to search one little document and it is impossible.

Footnote has improved their search function over the past year, but I would like to see it improved even more and with some increased speed. It is the slowest site I search with high speed internet. Footnote definitely needs some soundex and fuzzy search functions.

On ancestry I would love to see a "fuzzy search" that will pull Ballentine, Balentine, as well as Vallentine. Additionally if I could search the first letter by wildcard and then a series of letters I believe this would improve my searches. I also hope ancestry.com will consider keeping both the old and new search engines. I need the new one for Newspaper searches, but I much prefer the old engine for census, marriage and other records.

I would also like to see a better "card catalog" or index, which would assist me in locating the data bases I need to search at ancestry.com.

Sorry, but I could go on forever about search functions. Thanks for allowing me to vent.

Lynn said...

Two years ago I was looking for a family (since found) where the only information I really knew was the year and approximate month of their immigration; the mother was with them and the father may or may not have been with them; there were 6 year old twins, age 6, one named Peder and the other female; they were originally from Norway and they had a 4 year old brother. The mother, second twin and 4 year old had names that are spelled very differently every time I find something on them, so I didn't use them. And the surname turned out to be a farm name from the area in Norway they were from, but one I didn't know they had ever lived on --they used Paulson in Norway and the US; Moen on the ship.

It would have been nice to be able to put in more of the family in order to find them, like a mini-family tree... perhaps like Familysearch has now with parents, but with the addition of sibling information. It would have saved me months of searching and reading pages of manifests.

herzogm said...

HeritageQuestOnline has the big disadvantage of not allowing wildcards or Soundex BUT I love the ability to sort the results by category. If you can narrow your search by locality, you can then sort on age and quickly scan everyone in a five year span. Having one line results with many showing on a page also makes it easier to scan for the one you want. If you sort by county you can then quickly scan the ones that are possibilities without having to do a new search for each county.

Eileen said...

I would love to search the US census indexes by address like I can the UK census indexes at findmypast.

I like all your suggestions. I would like them to improve name searching so that it automatically searches on married and maiden name for females and if I enter Meisberger, I don't wish to have to page through 10,000 MCs and MACs before seeing another Meis...

Mostly, though I would like to be able to specify my order of preference for for the results. I would like to say primary surname, next residence or next birthplace. Depending what I am searching for I would want the results displayed to move my objective nearer to the front. This is becoming more and more important at content has increased at these site. These days a typical Ancestry.com search never nets me less than a couple thousand results usually it is closer to 30,000 and sometimes it really isn't feasible to reduce the numbers by filtering out databases.

Abba-Dad said...

1. Ancestry should allow wildcards with less than three characters.
2. I have yet to see a search engine allow you the use of before/after a certain date.
3. NewspaperArchive.com should get their act together in general.
4. Search engines should give the more common search fields first for specific databases.
5. You should be able to sort results of any search. How many times did you want to sort census results by birth date or location?
6. Pilot.FamilySearch and Footnote are on the right path.

So many more, but too late right now. Maybe I will write a post and let you know.

Geolover said...

I love the search engine on eB**. It searches millions of pages, actually implementing the search terms including the myriad boolean exclusionary items I use for some specific item-type searches. It is very fast, and if one uses the original search rather than the default 'beta' new search you can specify how the results are sorted in a usable way.

By contrast, the search engine interfaces on Ancestry.com and many other sites often disregard search parameters, do not implement date-bracketing, do not allow boolean exclusions to compensate for disregarding locations and date-bracketing. Furthermore, search results are not sortable in any meaningful way: present choices in the default "New Fuzzy Search Interface" are by date database was added, database popularity, title or number of name entries indexed in the database (which is humorously called 'relevance').

There have been many discussions of search-engine issues on Ancestry.com in its blog

http://blogs.ancestry.com/ancestry/category/ancestry-com/