* Fuzzy search

Requests that have been moved to the Wish List, or deemed to need no further action
Post Reply
User avatar
RogerF
Famous
Posts: 182
Joined: 26 Apr 2009 16:32
Family Historian: V6.2
Location: Oxfordshire, England
Contact:

Fuzzy search

Post by RogerF » 07 Dec 2010 17:02

As my database grows larger, I increasingly wish that FH supported some form of seaching which didn't demand precise values and specific records.

To illustrate the general form of what I'm after, imagine that I come across a census record for John Smith living with his sister Mary. I now have approximate dates and birthplaces, but no easy way of determining whether these people already exist in my database. What I need is a fuzzy seach, where relationships are (probably) stated, dates are approximate, spellings are loose and locations can be broadly sketched. Wildcards and Soundex probably play a part here.

What I DON'T want is a query tailored to this hypothetical problem. Rather, I anticipate a sort of visual parents and children mini-tree, where I can enter just as much data as I know, and FH replies by listed families which match my criteria.

http://www.fhug.org.uk/wishlist/wldispl ... lwlref=472

ID:4891
Roger Firth, using FH to research the FIRTHs of Lancashire and Yorkshire, and the residents of the market town where I live.

User avatar
Jane
Site Admin
Posts: 8441
Joined: 01 Nov 2002 15:00
Family Historian: V7
Location: Somerset, England
Contact:

Fuzzy search

Post by Jane » 07 Dec 2010 17:51

I realise you don't want to write a query each time, but you could fairly easily write a generic one to find families using soundex and range selections which could work to find what you want.
Jane
My Family History : My Photography "Knowledge is knowing that a tomato is a fruit. Wisdom is not putting it in a fruit salad."

User avatar
PeterR
Megastar
Posts: 1129
Joined: 10 Jul 2006 16:55
Family Historian: V7
Location: Northumberland, UK

Fuzzy search

Post by PeterR » 09 Dec 2010 23:24

I have managed to write a Query which will list families where two children in the same family (i.e. two siblings) have specified given names, e.g. 'John' and 'Mary', but I found it quite a chore, so I wonder if anyone has a better solution.  My Query firstly finds Individuals with the specified given name and then checks up to 20 siblings (children in the same family) for a match to the other specified given name; this involves a Row Filter Expression which is 1373 characters long.  There is also a Column Expression, 3556 characters long, which displays the given name of the first of the up to 20 children which matches the specified given name.  I can upload the query 'Individual Siblings' if anyone is interested, but I fear I may have missed a much simpler solution.
Peter Richmond (researching Richmond, Bulman, Martin, Driscoll, Baxter, Hall, Dales, Tyrer)

User avatar
RogerF
Famous
Posts: 182
Joined: 26 Apr 2009 16:32
Family Historian: V6.2
Location: Oxfordshire, England
Contact:

Fuzzy search

Post by RogerF » 10 Dec 2010 06:26

Thanks for your valiant effort, Peter - I can imagine it wasn't particularly straightforward. However, the 'two siblings' example was just that - a poor attempt to illustrate the idea of a completely generalised search. Let me try again.

Think of the FH Focus Window, showing the names, dates and birthplaces of grandparents, parents and children. Then imagine that each of those boxes is an empty, editable field. Into those fields, I can enter what facts I have -- names (first and/or surnames), date ranges, counties, etc, basically as little or as much as I know -- for one, two or more people. It could be three sisters, or husband and wife, or grandparent and grandson, or simply a widow... you get the idea. Surely FH can do a quicker job of fuzzy-matching the given information against my database than I can?
Roger Firth, using FH to research the FIRTHs of Lancashire and Yorkshire, and the residents of the market town where I live.

Post Reply