Duplicate Finder when creating profiles not very good

+26 votes
329 views
Harumph!

I just created a few profiles manually, and during the creation of all of them, NO possible matches came up. But now, doing some more poking around, I find that they're already here! UGH.

Such a drag.
WikiTree profile: John Seckinger
in WikiTree Tech by Jillaine Smith G2G6 Pilot (921k points)
I agree,
It has happend to me , not often but enough to be irritating.
After a while I learned just how bad the duplicate finder is. Now I spend a lot of time trying alternate spellings myself, and using google search with site=wikitree.com to search for duplicate profiles before creating a new one.
I want to second what Douglas has written! For a person born in the 1700s or earlier, I always try to look around WikiTree first to see if the person is already there. Many of my ancestors are Smiths, so I would look for a spouse with an unusual surname, or a sibling or child with an unusual first name, to be sure before I started that the person (and his whole family) wasn't already here.
Excellent practices, Douglas and Kay. I also look for dups, through the search facilites before creating new people.

I also search before adding a profile even though the duplicate finder when creating profiles still has save me from writing duplicates.

My tree was too huge to download gedcom. I research befozre I add them. Or I go out 2 or 3 generations to find the match.

3 Answers

+5 votes
Hi Jillaine,

That's odd. When I try to create a new John Seckinger, born 1865, the match comes up. Are you absolutely sure the matches section was empty?

Chris
by Chris Whitten G2G Astronaut (1.5m points)
100% certain on each and every one of the 4 duplicate profiles I just created? No. (But 90% certain; I'm pretty good at noticing that aspect.)

100% certain on at least two of them? Yes.

I do realize that one of them couldn't be helped because the person who created the earlier profile didn't know the maiden name of a wife (and I did), and she had used the married name instead. ALTHOUGH... I would think that since I did enter the married name when creating the duplicate profile of the wife, the dupe-finder SHOULD have looked for that, too. But maybe it's not that smart... yet.
If you let me know the IDs I can test what you entered.

Do you recall if any suggested matches were coming up, i.e. if you saw ones that didn't match?

Here are the ones I created this morning:

Seckinger-24 (merge has been proposed)

Cotter-678 (already merged away)

Seckinger-23 (merge has been proposed)

Schmitt-1681 (this is the one where the existing dupe -- Seckinger-8 -- was using the married name)

 

My recollection is that NONE of these four came up with a suggested match.
Chris, this just happened again.

I just created Andrews-8642 and Hough-1645, manually. Married couple.

During creation of each, no matches came up. But then when I went to create Hough's parents, I found them both!

I've proposed merges, but this shouldn't be happening, right?
Hi Jillaine,

Did you test entering the same information you entered for the people originally (without creating them) to see if they were found that time?

That would be a big clue to the source of the problem.

Thanks!

Chris

I have a pretty good idea why Jillaine's latest profiles didn't show up as matches.

  • The two Joseph Andrews profiles have birthdates more than 2 years apart, which means they wouldn't show as a match.
  • And the two profiles for Susannah have different spellings for her first name. The matching protocol now picks up many spelling variants for last names, but it's still rigid regarding first names, so it treats Susanna and Susannah as completely different names (a good reason to add variant spellings to the nicknames list).
Thank you so much, Ellen!

Other reasons for the duplicate checker to miss matches include:

  • LNAB of Unknown (which means that current last name and other last names won't be checked).
  • Space in the last name (last names with spaces in them, such as De Wit or Van Alen, do not get matched with variant spellings such as De Witt, Dewitt, Van Allen, and Vanalen).
Chris,

No, I had not previously looked for them outside of the creation-profile-comparison process. I had assumed that the search engine for both was the same. (Shouldn't it be?)

If the search feature for comparing a new profile against existing profiles doesn't work the same as the regular search, then why are we using it? Bottom line lesson for me here: use regular search first before creating a profile because the comparison-upon-creation search doesn't always work.
Jillaine, the basic code is the same, but the search page offers you options. As Ellen notes, some of the options on the automatic suggestions are limiting.
+3 votes
This has also happened several times for me also in the last few days. In fact just this morning it happened but I was fortunate enough to catch it in time without creating a duplicate. I then tried again using the exact data from a profile and it did not even catch it that way.
by Dale Byers G2G Astronaut (1.7m points)
Dale, what do you mean you were fortunate enough to catch it in time?

When I'm creating a profile and enter in the various data (names, dates, places), and NO suggested match comes up, I don't see how there's a way to "catch it in time". Please advise, because if there's something I'm missing I want to know about it.  Thanks!
Jillaine, It was one of the very rare times that by just entering the data I remembered the similarities to a profile from the same family I recently worked on and stopped to check. I then used a printout of the profile that was already on here and entered the exact same information, stopping short of the save when it did not register a match. I was fortunate in catching it this time but it has happened in the past when I did not catch it until I had already entered the profile and I just thought that I made the mistake.
The part that worries me is that the family's I am trying to clean up use the same names multiple times, in fact I found one case of the exact same name being used for at least 3 generations and in that case I also stopped short of saving just to check my information.
+5 votes

This is happening to me and it's very frustrating since I try my best to NOT create duplicate profiles. 

It seems to me that the profile finder isn't checking two things:

  1. It is NOT checking married name. When I put in a woman's name, such as Mary, LNAB: Rodman, Current Name: Carpenter. The "Carpenter" results are not coming up. What is coming up is a bunch of woman who's LNAB is NOT Rodman. This is very frustrating and has caused me to miss dups when someone has created a profile for Mary (Unknown) Carpenter. This has happened to me a few times this month.
  2. It is NOT showing profiles without birth dates. When I put in a name like Hannah (Rodman) Mitchell, DOB 20 May 1749, no profiles were found. As soon as I tried to add her spouse William Mitchell, I found him immediately with a profile for a Hannah (Rodman) Mitchell with no DOB. This has happened to me at least a few times this month.
  3. Editing and adding a 3rd scenario, typos and different spellings. Clarke vs Clark, Katharine vs Katherine, these kinds of spelling differences are not found in the profile finder, but are in the person search.

These aspects of the finding algorithm really need to be fixed as soon as possible as even someone that is diligant, like me, ends up creating five or more dups in a month.

by Allison Mackler G2G6 Mach 6 (64.5k points)
edited by Allison Mackler
I concur.

Related questions

+17 votes
1 answer
+43 votes
9 answers
+7 votes
1 answer
136 views asked Mar 9, 2015 in WikiTree Tech by Antonia Reuvers G2G6 Mach 1 (19.4k points)

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright

...