Searching by Value List?

[Sunnylew]

I'm not sure if this is the right place to be asking this question, but as far as I can manage to figure my problem out, I think I must need to leverage Value Lists.

I am doing a genealogy one name study which is source/event based. I know that genealogy dtabases are the deep end, but have managed to work something out to my satisfaction so far.

However, I am stuck when I come to sort and finding - I want the database to know that John, Johannes, Jo:, Jhon, Jonothan etc are all exactly the same name.

And the same goes for surnames, placenames, counties, occupations, titles etc.

I can see how I can make a "Normalise" table create itself, where I am asked for the normalised version of, say Johannes when I enter the name. If Johannes has already been entered, I would have previously entered the name "John" and it would not ask, and eventually all instances would have a related normalised version in the "Normalise" table.

I can create a value list from this table, and sort according to normalised versions (I think). But what do I do when I want to find every John in a Find situation? And what if I wish to find every John/Jo:/Johannes Whyte/White/Wight?

Can I use value lists somehow, or would this need some form of tricky scripting?

If anyone could help me even conceptualise the flow of what would be needed here it would help me no end.

Lewis

[Peter (duksis3)]

As I understand - you want some "guess" or database must guess what you mean or what word mean.

Databases are strict and "guess" isn't OK.

My opinion is that you need more carefully decide and think about database build.

Task is solvable but solution design take some time.

[Sunnylew]

I definitely don't want a guess

Genealogy often uses algorithms such as Soundex and Simplex which try to guess on what mistakes people make in spelling and hearing to create other options for spelling. I find that these results are often misleading, and are no help when it's a name in latin versus one in English.

I want to dictate what each thing to be treated the same is. I probably didn't explain myself well :

I have already developed a small script that will add a new record to a "Normalise" Table for every name/term/place that is used. So I end up with a 2 field table which under "Original Term" has Jhon/Johannes/Jo:/Jonothan , and in the "Normalise" column has "John" beside each.

What I want to do is have my database treat Jhon and Johannes and Jo: and Jonothan as exactly the same term, because however they are spellt, they could all be exactly the same person.

As I am dealing with evidence based sources, I can't change "Jo:" to "John" when I'm entering the record, as that would corrupt the validity of the original source, even though I know it is simply shorthand for "John".

I'm hoping that Value lists created from the fields in my "Normalise" table can do the job.

Otherwise, all I can think of is perhaps having a mirror layout of each record created with the normalised terms replacing the original. I could then construct my own Find script which searches the mirrors but returns a display of the original record.

Does any of this sound like the right direction to take?

Lewis

Edited November 8, 2009 by Guest

[Peter (duksis3)]

OK - now I understand a bit better

I suppose that you need to look on parent - child records representation or hierachy representation.

"Normalized" term would be parent and others - child.

Take a look on post here .

You need to go forward and create several hierarchies and combine them together via relationships (with FM10 you can duplicate tables via self- import).

If I have some free time I'll make some sample solution (from my point of view) and upload it here.

Peter

[Peter (duksis3)]

I have made sample relationship file.

2 varints - 1st is based on first letter, 2nd is based on "normalized" term field.

If proposed is what you want :laugh:

Normalized.zip

[Sunnylew]

Thanks Peter.

I've set something up now based on your ideas which works quite well.

I want it to function invisibly, so when I am happy with the layout I am using I will duplicate it and have the normalised fields in place of the regular fields so that a script can perform the find there, then return to the origianl layout.

It's strange how complicated everything sounds until you've actually put it together yourself! :)

[Peter (duksis3)]

OK.

Try List function from relationship field too. And try to put result as Tooltip (calculation) - FM10 must bee in operation.

Then you must see variants after moving mouse over field. Tooltip is delayed for a while but sometimes it is very good tool.

Good luck!