Very Frustrated with GEDCOM

Question

I uploaded my GED file with the intent of adding my family tree to Wikitree. Because of the number of entries and the prevalence of common family names there are thousands of "matches" (only a tiny proportion of which have any significant commonality with the imported person) to clear before I can actually add any of the people. This makes this site pretty much unusable for me. Just a little feedback on my way out the door. My family tree of 1200 researched and well-documented names won't be added to your database. I've already spent hours doing this, but I'm not going to continue. It's just too much work that I've already done once.

Comments

Dear Anonymous- sorry to "lose" you when you have just begun with WikiTree. As you undoubtedly have learned we are creating one large family tree with a single profile per person...it is possible that your "well researched" names are already included here, but additional sourcing is always wanted- hope to see you again soon, we'll miss you in your absence. (Ialways tell my friends that we are FREE, FRIENDLY, and FAST to learn...but I guess you already know that!)

---

No need for such condescension, and I definitely will not recommend to friends or family.

I went through about 1000 names found only a few matches and still couldn't even import my brother. Most of them only matched on one particular, the last name, not first name, birth date, death date, country of birth, father's name, mother's name. Frankly it was just a giant pile of crap to wade through. The frustration finally overwhelmed me. Good luck with your project, but you'll do whatever you do without any help from me.

---

Condescension? If it's my comment you are referring to, I definitely apologize, Wilson. It was unintentional on my part...though I'm not quite sure if it's my comment that is being referenced. Anyway, we really are a friendly, informed group...an hope that you will be comfortable using this site. Sorry.

---

My point is that from the perspective of someone who has done thousands of hours of research and building out of a family tree, your onboarding process is horribly, horribly broken.

Do with that information what you will.

---

I agree with Wilson that trying to add new names to WikiTree by uploading a GEDCOM is just not worth the effort.

I also agree that protestations that we are sorry to lose people like Wilson ring hollow. For a very good reason: we are not sorry.

"Al wie met ons mee wil gaan, Die moet onze manieren verstaan". Those who would go with us, must understand our ways.

---

Your loss, not mine. This site has given me nothing but frustration.

---

Apparently Anonymous confirmed some of the matches in his Gedcompare, but didn't finish that process, and was not thorough about marking rejections. Probably the reason he couldn't add his brother is that there were suggested matches for Anonymous's own profile or other members of the immediate family that hadn't been rejected yet.

Uses of Gedcompare take notice: You can't just mark the matches, you also have to mark the rejections.

---

This is not a correct description of either my experience or how the process works. Read the help file.

---

I understand why the process for importing GEDCOMs is what it is for WikiTree but I also think there are some simple UI improvements that could make the processs less painful.

1) There has to be a better process for sorting through matches — even just having a view that only shows Suggested Matches would make the process so much better.  

2) I should be able to reject all matches to a profile with one click, and even if I have to reject multiple matches, I should be able to do that on one screen without having to go back to the profile over and over again.

Answer 1

Personally, I prefer to enter mine one at a time, and I have far more than 1200 to do - I hope I live long enough! I've come across the mess some gedcoms leave, and it sounds like you wouldn't have time to make sure each profile was the best it could be.

Why not try a smaller gedcom with just one line of direct ancestors. For a little effort you will have set the groundwork for great cousin bait.

Comments

Gedcompare is an improvement over manual profile creation.

In Gedcompare, we add profiles one at a time, but the system fills in the blanks on the profile form for us (using information copied from the GEDCOM). Therefore, instead of typing information into the form when we create a profile, we only need to edit the information entered onto the form.

Try it, you just may like it! But do start with a small GEDCOM.

Answer 2

Let's face it, WikiTree is not the ideal platform for everybody. If you want to minimize duplicates you have to strike a balance between being too strict in your search parameters (and allow the creation of more duplicates) and being too loose (and create more work for people wanting to create profiles). I suppose WikiTree has come down on the side of expecting their members to do a little more work. That does not suit everybody and we as WikiTree members have to accept that, too.

Comments

Yes. I determined that Wikitree is not the place for me, so I left. But you should be concerned that happened because I have lots of valuable unique data that doesn't already exist in your system and was trying in good faith to add it. I have a completely built out tree based on the genealogical reference "The Clarks of Pennsylvania and Allied Families" almost none of which exists in wikitree. Likewise I have a thoroughly researched and annotated autobiography of one Reverend John Clapp Baker which exists no-where else on the internet and my intent was to upload it here.

I am not a neophyte with online communities, nor wiki projects. In fact, I've been an online community manager since the 80s and I'm meta at Wikimedia. Again, if you are trying to attract people who have already done a lot of work on their genealogy then your (apparently newish) GEDCOM process is horribly broken.

I wish you could somehow experience what I have experienced in dealing with the GEDCOM compare for hour after grueling hour of people who are clearly, obviously, not matches, and knowing that you still have many dozens of hours of clicking and clicking and clicking to go before you will be able to import any of your already thoroughly researched work.

If I were to mention a few things that could be fixed that would have made a difference they are:
1) When you unmatch in the compare, instead of dumping you back into the main screen it goes onto the next person identified for compare. (This would have saved me a screen resize, scroll, click, wait 4 seconds for the compare popup to load, and all the interruption of focus that comes with that.)
2) Do not bring up matches that only match one key point (last name) If it's not the same first name, and it's not the same birth year, and it's not the same country of birth, it's not the same person.
3) Be cognizant of maiden names vs married names. In my GED file all the names are birth names. This means that for every single woman in my tree, more than half of the matches were obviously wrong.
4) Design your interfaces so they work with a tablet. Your fonts are too small and your input dialogues are not dynamically resizable.
5) Allow new contributors to add people from their GEDCOM file that are closely related to them regardless of whether they have completely matched the entire file.
6) Once a person has done 100+ comparisons, they get it. They should be able to add people who are directly related to them who they have already gone through the process of checking for matches.

Good luck with your project.

---

Too bad Wilson didn't ask for advice before he got so frustrated. He could have added his brother to WikiTree if only he had reviewed all of the matches for members of the immediate family.

---

I just wanted to say Thank You, for all the time you took to offer constructive suggestions.  I believe about half of yours are already on the ToDo list.  Improving WikiTree is an iterative process, and there's quite a ways to go yet.  Putting in as much effort as you did does show you gave it a fair chance, and respectfully suggested what would have worked better for you, valuable feedback.  I suspect we lose others who leave quietly (or at least cuss quietly as they go!)

---

I actually have multiple GEDcom files. Since I am a geek I can use notepad to break it up.  I keep a copy of the family and source information with every copy. That way I am only looking at 50-150 at a time.  Also, if I have to modify the GEDcom and reupload it I will put a \ on the 0 INDI line and it is ignored during the import.

 

0 @I500084@ INDI Hester Ann \Minton-999 in Wiki.  Some of this information needs  to be put there and sourced brilliantly.
0 @I9907@ INDI \On Wiki
0 @I8509@ INDI \TryMe
0 @I12621@ INDI \TryMe Thomas H

Answer 3

I too am sorry to see you go as well,  I understand how daunting it appears when you load a gedcom file with over 1,000 names and  you see a lot of Suggested Matches.  Especially if the names in your file are considered common.  I have Johnson, Williams, Smith, Yoho in my family and I tell you, it's not easy.  OK, Yoho isn't a common name, but I just like spelling it.

Why are so many of the Suggested Matches not even close in your view, as well as mine.  Because many names already entered into WikiTree may not have been as well researched as yours.  You may find a duplicate with erroneous data that needs to be addressed is one reason.  Also, many names entered have very little statistical data which is really needed to perform a more exacting search. And lastly, the search uses SOUNDEX, a phonetic coding system intended to suppress spelling variations, used especially to encode surnames for the linkage of medical and other records.

Think of it this way, if I had access to your historical files, would you want me to start adding names without making sure I verified them first.

When using the GEDCOMpare, WikiTree is not asking you to completely redo the research you have already performed.  It's just asking you if someone has already entered the name you wish to add. Having one profile per person is a fundamental part of WikiTree's Single Global Family Tree mission.

Christine had a good suggestion that if you could start with a smaller file size, it is much easier to use the GEDCOMpare and quicker. Especially with common names, trust me on that one. LOL  I am not sure what program you used to create your gedcom file, but could it possibly have a means of breaking it done by a single branch or range.

WikiTree only grows as a viable genealogy site through the addition of well researched data such as yours.  I hope you reconsider.

Comments

I think you are on to something here. GEDcompare should not offer you a match to a profile that has very little data. But how to detect that programmatically?

---

Great Sam.

And as you progress through your names, it gets easier and easier.

And now that you have experienced the joy of Suggested Matches (LOL), you understand why Aunt Liza may have to wait till later when you finally find the elusive record or two. Though gedcoms are my predominant method of adding to WikiTree, I still have to add a bunch manually after I get better records as I have more than a few Aunt Liza's in my tree as well.

And kudos on sniffing out the fix for Mary Cate.

All the best,

LJ

---

Well Dirk,

The search is based only on the data fields and it would be easy to program it to skip empty data.

But I have come across more than a few Profiles where the information in the data fields is not correctly filled or empty, but is available in their Biography and I have been able to Confirm or Reject based on that.

Creating a search program to parse through the Biographies would be nigh on impossible.  Could it be done, yes.  But this is WikiTree manned by volunteers and not the NSA.  And again, WikiTree is international in scope, a parsing program would have to incorporate the languages of all who participate.

Fortunately, WikiTree has folks who are called Sourcerers and Data Doctors who do go through the system and try to correct the incomplete or bad Profiles.  Kudos to them!  Part of being an Open Site is having to take the good with the bad.  As someone once said, "This would be a great world, if it wasn't for all the darn people".LOL

---

Personally, I did think of a couple ideas for improvement.

Rejecting a match means you have to go back a few steps to get to the next suggested match.  I would be easier for the end user if when a match is rejected to go to the next suggested comparison.  This would clearly speed the process up.

It seems that profiles might be assigned a "quality score" based on how many sources exist, have many references, how much data is in the profile.  Then the end user might be able to filter out suggested matches of very low quality scores.  This ls significantly reduce the number of false positive.  Currently GEDCompare has high sensitivity and low specificity.  This leads to a lot of frustration of someone like Wilson who clearly has a done a monster level of research and cares a lot.  Losing people like Wilson is costly to the project.

---

LOL Sam, that Reject and back to the Table has ticked me off more than few times too.  I've been meaning to add that to this page.  Please go ahead and add a reply stating this at the bottom of the url. Get your feet wet on reporting issues.

For suggestions on improvement of GEDCOMpare: :  https://www.wikitree.com/g2g/461205/have-you-tried-gedcompare-2-0 

The Quality Score is a bit hard. As the area where the sources reside is a text area where anything can be added, I don't think it's feasible for this to be done.  Computers are good at true or false, but maybe's, not so good.  Though WikiTree has does have standards and styles on Sources, but like the data fields, folks don't always follow good practices.  It still requires human intervention to decide if a source is good or bad.

---

Dear LJ Russell:

Hey I have some programming experience and have helped run an open source development project previously so i understand this very well.  It's just that a profile that has no links to other profiles, no parents, no children, no siblings, source textbox unchanged from default, personal stories unchanged from default. This type of profile made up about 2/3 of the suspected matches.  I would think that the algorithm could be changed to detect that there is just a name, suspected birth and a suspected birth location and all the rest of the data is "null".

Just counting the number of characters in the sources textbox would be useful.  The default boilerplate has about 120 characters. Each additional source adds 50-120 characters.  A source textbox with (800 characters less 120 for default) likely contains 5-10 sources.

Personal Stories is likely going to contain 55 characters per line.  The default boilerplate is about 50 characters.  The amount of useful data would be the total character count less 55.

Sam Bowen

---

Hmmm Sam, may be on to something there.  But I am not part of the programming group.  I just help folks get it to run, not make it run.  ;)

Your suggestion needs to go to go here to insure the programmers see it.  For suggestions on improvement of GEDCOMpare: :  https://www.wikitree.com/g2g/461205/have-you-tried-gedcompare-2-0 

Think of that as the Technical Support Page for the GEDCOMpare programmers.  A posting on that page will get to the folks that matter.

I know they are very receptive to any suggestions as the GEDCOMpare was just instituted last fall and is still considered a work in progress.

LJ

---

Hi, LJ. I’ve just joined Wikitree (April). So I’m a little late in responding to this thread. I use Famiky Tree Maker (2014.1). My database has over 67,000 names it it, and I’m not about to try to upload my gedcom. I tried the link for the older versions of FTM to break up the gedcom, but apparently it’s a dead link.

It would make it easier to break up my gedcom/database, but I rather enjoy entering one person at a time. I get to focusing on one family and start forgetting things I once knew off the top of my head about other families. (For example, in entering some of my Underwood collaterals I “rediscovered” that one line of them are double cousins: they are also Lawing descendants, as am I.) Of course, I’m retired, so I’ve got the time and the patience.

But not everyone is like me (thank goodness!), and I can understand Wilson’s frustration. Been there, done that. One of the things I know I wouldn’t have patience for merging the hundreds (maybe thousands) of matches. That what happens when one finds that golden connection to European nobility and royalty. So, I’ll just keep adding one at a time and request merges or add parents when  they pop up.

I’ve already had some wonderful help here on WikiTree, and hope I can be as helpful. Any help someone could give me in splitting up my gedcom would be greatly appreciated.

---

Hi Pip,

Better late than never as the old saw goes.  And welcome to the club!

Wish I could help with the FTM, but I am  not a user.  However, I did run across a program called Legacy Family Tree ( https://legacyfamilytree.com/ ) which does have the means to break down your tree by surname, date range, etc.  You may want to take a look at it.  It cost about $50.00  I have only played with it a little, but what I saw intrigued me.

I plan on putting my tree(s) into it to have complete tree of my own as well as on WIkiTree.  I haven't yet as I am slowly going through all of my present WikiTree Profiles and cleaning up those I first added and re verifying data and adding new data as I update them, both on WIkiTree and my own trees.  I had to do this as there is a 5,000 name limit on the size of your Watch List which I hit and I have more than that to add.  So I am Orphaning those related only through marriage and keeping only blood kin and their spouses.  I estimate at my current rate, I may get done with this about September. This 5,000 name limit is something you may want to keep in mind what with 67,000 names...WOOF.  I thought my 15,000 plus was bad.  LOL

Well good luck on your endeavor and take a look at that program to see if it needs your needs.

All the best,

LJ

---

Morning (here), LJ,

Thanks for the welcome. I’m having a blast with Wikitree.

I have to admit that 15,000+ of the names in my database came from a merge with another researcher who had, in collaboration with others, created a HUGH tree of descendants of Andreas Killian of Germany and NC. So I can’t take credit for nearly all of what I have.

Thanks for the advice! I’ll check it out, and go through the FTM manual to see if I can find out how to make a smaller gedcom of branches of my family and see if I can upload that. While I enjoy entering one at a time, it would be a lot easier, especially with the interconnectness of my various families.

Thanks again, LJ!

Kind regards,

Pip