Guidelines for working Database errors

Question

In case you have not seen this, we have a new Database Errors Project.

http://www.wikitree.com/wiki/Space:Database_Errors_Project/2016_05_01

I have a couple of concerns/questions for everyone involved, and if I am totally clueless here or missing a point...let me know.

1) The list you get when you put in your grandfather or whoever, is every connected profiles, not just the ones that you manage....I think that needs clarification, and that means that even though it is on "your" error list, you need to collaborate with the profile manager, especially when it has to do with dates, mother is too young/old...those kinds of things.

2) For the project participants, I believe it should be the same process.  A couple of people I have conversed with, thought that if it is on the list, and they work the list....they have the "right" to "fix" things.   I say, we are still a collaborative site, even when we find errors, or work a project.

3) The changes need to have comments made, just because they are on a "list" should not mean that you do not have to explain your change.

Did I misunderstand the mission of this project?   Am I incorrect in the assumptions above?   I read through the project pages and it sure sounds like..."a description of the kind of error and how to fix it."   Perhaps I am too concerned, but, there are a lot of people who are excited that they get to "fix profiles".

Comments

Gilly, When you hover on the underlined flag text it says it means spam or inappropriate comment. Just trying stay off the naughty list here.

---

I think we need to give people a chance to do their own thing also, ie tell them there's an error message, what it consists of, and let them take care of it.  It's not like there aren't enormous quantities of  orphan profiles to get sorted out first, if you are looking at doing more than your own family tree.  Granted you want to get correct data everywhere, but just from the number of errors cited above, it's not a short job.  

I have been dealing with a very upset lady recently after I proposed a merge of profiles, which were very much the same man.  Seems she didn't get the picture of how Wikitree works, even though she says she's been on here for a long time.  I could have done without the upset she threw my way frankly.  So, for profiles which are not managed by me, I've been putting messages on them unless the error was something like USA prior to its formation etc, and then I put a note on why I changed it in the box for that.

But for my own tree, I'm actually finding this new tool very nice to use.  Catching oddball things like wrong brother's birth date entered (namesakes), or wrong century typo, etc.  I'm like the kid with the new toy.  :D

---

OH good heavens and I've red flagged other comments - one or two. Thank goodness I've see I can unflag. My profuse apologies for anyone I've flagged - I thought it meant your comment was terrific and now I find I've flagged you for Spam oops!

---

As far as commas in name fields, they get used in other last name box as a standard measure, since folks around here had multiple choices sometimes, and the program does NOT differentiate them if there is no comma between them, in fact the program will not find something if there is no comma.  For example, the name Béliveau dit Beaugrand (invented here) is the format for many names, Béliveau being the LNAB and dit Beaugrand being an aka (dit translates as 'called': Joe Blow called Smith).  Do a search for just Béliveau, or just Beaugrand, if the name has been entered in one stretch as above, the program will NOT find the guy.

So, commas indeed should be in those fields when required.

---

Mikey I apologise profusely. I made an awful mistake. I thought that a Red Flag meant your comment was really good but now I find it means it is spam. I removed the Red Flag I added to your comment as soon as I was told what it meant. I really am very sorry. Your comment was an excellent one.

---

Thanks Gilly :-) I miss a "like" button too in these G2G feeds ...

---

Thanks for bringing this up, Robin. It looks like the Project Page for DB Errors has incorporated many of the suggestions from this thread. Thank you to those who are helping shape best practices for the project!

 

---

I agree 100% Danielle

---

Robin, thank you for addressing this issue #3.  I agree that the changes need to have comments made.  It certainly helps in the long run, even though it doesn't always save what you write, but, it's close enough.  Any and all changes need to be explained and when not, then you can't go any further and forced to but in a kind and loving way.  To those who need help always ask, WikiTree community are here to help.  Great topic Robin!

---

To be clear, where there is conflicting data, the conflicts should be addressed in the body of the text on the profile, not in the explain your changes field, as no one sees that information unless they are looking for something specific.

Answer 1

Hi All,

Here is my personal code of conduct that I have been following when fixing database errors, I have had no complaints as yet and lots of thank you's so maybe this is a way of working  we would like to adopt for the project

My code of conduct when fixing database errors

- With very easy obvious errors such as incorrect gender, gender missing, there is no need to contact the profile manager for an unlocked profile, plus the change log shows exactly what happened.

- Be clear and friendly when advising of errors, errors happen to everyone! be careful  to try to not make people feel like there in the naughty corner.

- Post a message on the public comments to say if you have contacted the person,  so that no one else also contacts them. - NB - I think public comments should  be ok to be deleted once they are finished with and think we should have a debate about this soon.

- If you do make changes such as changing a unknown birth location to a place, make sure  a source is added to the record so the profile manager can have a look.

- Make sure all correspondance has the url of the error thread so they can look at it,  this also helps people see there not being personally targeted

- If you see the same name coming up regularly or if the person is pretty well known on g2g drop them a more general have you seen this type message and leave them to it for a little while  (Ales might be able to help with this by adding in the project manager name to the error list's, but not sure if that could make it too much of a league table)

- Ensure if the above is the case the profile manage is aware that after getting a list of their errors off they can hide them for 30 days while they are working on them, so there would be no need to send messages or put a comment on for every profile.

- If people come back to you for advice, further information help them in a timely manner or direct them to a person who can help. I also advise to add announcements to their tag to find out more about these types of projects.

- If you’re working on a particular line from the error sheet, add a note on the project page so no one else will also look at it that way.

- If you put through a merge then you and the profile manager are responsibly for taking the duplicate notes out, I usually give the profile manager a few weeks to do that, if they then haven't I do it.

Comments

Profile manager is not part of database dump, so I can not help you here.

I totally agree with you. Can I put this text with minor changes to Project page?

---

This is an excellent answer.

---

I would modify the first line

With very easy obvious errors such as incorrect gender, gender missing, there is no need to contact the profile manager for an unlocked profile, the change log shows exactly what happened, but a simple message in the "Explain your changes" box saves having to open a profile to check for the change.

---

Hi Ales, You are welcome to use any part of my post you would like to, but maybe its worth waiting a little longer to see what more experienced wiki treer's (hmm whats a collective name for wiki historians') think first :)

Thank you Jillaine :)

---

Excellent answer Paula!

---

I used the Database Errors Project by inserting a couple of my direct ancestors WikiTree ID's and pulled up the errors on profiles related to these ancestors.  To profiles I am not PM on I left notes when it was more than taking out USA in locations that were too young at the time or just adding a gender.

I posted notes on profiles pointed at either the parents date of death precedes the birth of the child or the parent was not old enough to be the  child's parent, and where the child listed was born to parents who were born 200 years before them.

I have gotten several responses from the PM's that astounded me;

See the following responses;

Cheryl, Re: Your comment on Mary (Durham) Dodson's profile: " Abraham Dodson, Joshua Dodson Sr. and Elisha Dodson Sr. were all born after Mary Dodson died.They are probably linked to wrong parents. Please review profiles and correct." NONE of these individuals appear in any of my records. They were apparently added by a merge. I have pretty well ceased to participate in this fiasco & usually just approve any merge that has any semblance of possibility. It looks like many on here are pushing the "I wish it were so, therefore it must be" agenda for colonial times. Frankly, I am more interested in the gaps in my trees AFTER 1800.

and

I also believe that his parents are incorrect. In the biography of James, it states his mother was Mary Ann Hitt not Elizabeth. Several generations of the Rector's named one of their sons Jacob. So most likely it was a Jacob born earlier that married a Mary Ann Hitt. But, I have not proof If you look at the history a Dennis Hayes was the person who updated the link in June 2015 to add the parents. As a member of the Wikitree community you can update the profile. While, I am the profile manager of one of Jacob Rector, I only adopted the profile to update it with some additional information.

Am I mistaken or are the PM's wanting me to do the research and corrections for them?  What protocol do we have for situations like this?  Since I pointed out the discrepancies to the PM's, is it assumed that I should be the one to also do the research and corrections?

If I get post on errors I do my best to find out how they happened and find corrective sources or if typo fix myself, I do not expect the person posting to do it for me.

How should this be handled?  Don't want to step on toes but mine are getting bruised trying to do the right thing.

---

And if you propose to merge obvious duplicates, they'll talk like it's your job to make the choices and resolve the discrepancies.  But the PM is supposed to "lead the collaboration".

---

Cheryl, I'd be tempted to respond to some of these with, "if you are no longer interested in working on this profile, you might consider removing yourself as profile manager or downgrading to watchlist. See the help page for profile managers: http://www.wikitree.com/wiki/Profile_Manager "

But you also raise a great question about the db_errors project (and even wikitree over all)-- what is our role as project (wikitree) members? It might depend on which error code you're working on. I work on 902-- open profiles with all empty data fields and most wholly disconnected from the tree. If the profile is orphaned, I fix the problem myself. If there is a profile manager, I leave a comment on the profile asking them if they can at least estimate some dates and connect the person to another person. The responses I've received have mostly been positive and thankful. A few have approached the type you're receiving but usually giving me permission to do with the profile what I want. If I have time, I'll try to find some dates and connections or merge the person away.

If you don't want to do the research or have time, I encourage you to let such responses roll off your back (your toes?). Or wear stronger boots.

There are many of us who appreciate what you're doing.

---

Jillaine, I saved your wording to use because my wording would probably be "You've got a choice: Lead, Follow or Get Out of the way."

---

I'd like another few obvious errors I see no reason to contact anyone before correcting

1. Abbreviated or wrongly spelled city or state names.  I've changed NC, Ilinois, and Commecticut more than once. Once I changed Kazoo to Kalamazoo.

2. Obvious reversals.  The unusual name is Nleson, look at the profile it is only cited as Nelson.

3. Punctuation. Comma instead of period. Colon instead of semi-colon. Double periods or commas.

4. Unbalanced  ==  or [[.

Answer 2

Thanks Robin, You are correct.  

I think we went a bit too fast on this and did not have a lot of information or did not readily find it as we learned of it.

I am guilty of finding all these great errors on "My Profiles" and got drunk on fixing them, I did discover that I was running into a lot of profiles that were indirectly related to me after I had made some corrections without comment. 

All minor corrections but I can see how this would drive another profile manager crazy who suddenly saw somebody make a host of changes on profiles we shared and had no clue because there was no comment.

My apologies. Thanks for the smack in the head, it was deserved.

Comments

While I agree that adding a comment is useful when making a change, it's not as if a Profile Manager is completely in the dark about what happened. They can always look at the Changes tab to see the exact changes that were made.

---

Yes, Eric, you are correct.

Unfortunately, that is exactly what Robin did, over and over and over because she wants her profiles to be accurate.  One or two profiles are not so bothersome, but what the Error tool did, was give us access to thousands of errors (yes thousands) in one big swoop.

So, if some did like like I did, and made about 40 corrections in 30 minutes and because they overlap with Robin, she suddenly saw massive corrections to her profiles and wonder what the heck was going on and started checking each one.  

Had I commented, "deleted USA from date" she would have been clued in.

So you are technically correct, but I do not want Robin pissed at me.  I need and enjoy her help.

---

Ahhhh. Yes, that makes sense. ;-)

---

At some point, I hope to see the error checking system integrated into the main WikiTree interface so that errors are more visible and error correction more directly auditable. As part of the integration, the distinction needs to be made between errors in the profiles you directly manage, errors in the profiles other people directly manage, and errors in orphaned profiles as well as the permission/privacy system.

Comments are a good stop gap measure.

---

That is also my wish, for user to see all errors of his managed profiles. But since I don't have access to manager of a profile I can not do this without Chris's help.

If anyone knows a way to get list of managed profiles let me know.

---

Integration would be great and I'm sure it will happen in some form down the road.

Honestly, once we integrate, progress will probably slow down dramatically, since we have so few internal development resources and so much we'd like to do. Right now Aleš is running so far and so fast, I hate to slow him down.

Aleš, the database dump could include a manager ID. Only one, though, so if there are multiple managers you wouldn't get them.

I think Trusted Lists are better identifiers of the interested parties. The manager doesn't own the profile. The way to use Trusted Lists would be to somehow use the getWatchlist function of the API.

Answer 3

Robin, I think you have a good point.  I have "fixed" a bunch of profiles with simple, such as no or wrong gender or with USA before 1776.  I did put something in the Changes line, but I didn't feel that it was necessary, for these simple changes, to contact the profile manager.  For other, more serious problems, I have contacted a bunch of profile managers.

The concerns I have are with early profiles with oddball names, siblings with the same name, estimated dates, etc.  Many times these are very complex profiles. Without good researching and sourcing skills, there could be a lot of damage done in the name of "fixing" problems.  The MINIMUM would be to work with any project that might cover the profile and with the profile managers.

Many pre-1500 profiles have estimated dates, and those dates may show up as database errors. Profiles of persons born before 1500 should be handled by people with pre-1500 certification, and there should be no special dispensation to anyone in awarding that certification just because they are part of the Database Errors Project.

Just being a member of the Database Errors Project is not sufficient reason to abrogate our Honor Code: we collaborate.

Comments

In full agreement Vic!

---

Pre 1500 profiles can only be edited by certified people. That is the reason, to have separate list for that era, so uncertified people don't even try to check them.

---

The {{Dutch_Cape_Colony}} project is from 1652 (but with a scope that starts pre-1500) to 1806. There are no limits for people who'd like to edit the post-1700 part of the project (or even create manually duplicate pre-1700); the biggest looming issue is the exponentially growing amount of nineteenth century profiles that have dated and contradictory secondary sources - because of historical reasons the primary records for this period can at times be much more difficult to source than 18th century profiles ... merely "correcting" "errors" without regard for the context can be evenly un-validatiing .... But I agree - this is an amazing initiative and I'm going to spend a big deal of my weekend using this error-database correcting profiles ...

I have a heart for WikiTree even despite it's issues ...

Answer 4

Unlike most on WikiTree, I work entire "lines" from present all the way back into pre-1500 without stopping until the end.  One thing I know after at least two and a half years of this method is that you still have to post convincing proof (we use original document images more often than not) and wait for consensus and discussion.

One reason is that even if you are clearly right - the profile managers need to understand how and why. Just because you have a flawless source does not mean the PMs will understand it or its implications. (For example, imagine a  family where for five generations there are at least three John Johnsons each and you have a will!!!)

Just because the dates don't "work" does not mean that is the problem.  The problem could be a wrong connection or a bad merge or other things. The people that manage the profile need to be engaged to be emotionally invested in preventing future errors.  If they are part of the process they understand what to watch for and are excellent at it!! Also, a database is not a person. A profile is or was and the managers are   So that activates the honor code.  

I have worked in tandem with Eric Daly on Southern Colonies profiles in this      manner constantly so I have someone doing exellent indepth research before I make changes. He can spend 40 hours a week researching one error and I am in charge of communicating with any interested parties.  In older profiles, all intereted parties are not immediately evident!! Still, despite that and a very solid familiarity of the lines I work (most are also my own), I make mistakes!! So instead of saying we are trying to "fix" things. I report we are working lines adding sources and writing bios.  Even still, understandably people can be concerned at first. Taking the time to communicate with them builds necessary trust and shows respect for their work.

I am not setting myself or my work up as an example of the right way to do things or anything at all of that nature.  I just (without our new tool) have worked entire lines long enough to have one thing I can offer - much care should be taken! A database is an entire genealogical history!!!

One last thing - I would never correct profiles from an era or region I did not have a very strong knowledge of when there is another project that is available to do that!! Oh yes - are you going to contact all involved projects?     

There is no one last thing .... this list goes on and on!!!  I do applaud you for your work!  I really do!  Most errors I have seen involved tangled lines. Bad dates may or may not be involved but they are a clue. I would hate to think an entire project was making sure dates worked without taking time to research the family to determine if possible cousins were confused.

Thanks again for your hard work! Thanks for taking time to read this. And thanks to all of the members of WikiTree who have collaborated with me and Eric, as well as Cheryl, Mags, Mary, Terry, Robin, Liz, Allan and many others who work these same profiles in making sure our Southern lines are solid and represent the people we are today as Southerners!!

Comments

Someone needs to remove the red flag. I think someone must have used it incorrectly. Paula's answer is wholly appropriate.

---

Well said Paula,

After working with you on projects and profiles, I find your way of making the PMs understand is the best way,  We are here to work together and produce the best profiles we can to honor our families and ancestors with the most accurate information available.

Answer 5

But most of the errors on the list isn't it just lack of validation when adding a Gedcom file or a profile..... and many could been corrected by a bot.....

i share your point of view that we should communicate but i can also see that this person doesn't have a gender ==>  it's a no problem for me to change it plus we can always use Wiki standard restore....

i add a section == Research Notes == and a {{db_errors}} to make life easier for other people to use this magic tool...

Great tool and it makes us look on the WikiTree data in a new way..... i hope we get more tools like this and also that Wikitree starts thinking about how we make Wikitree easier to work with. Today we use to much Free Text when we should add data as data e.g. a location a person lived in should not be stored as just text it should be a timestamp and a location that a software could read and use..... 

Comments

Example of me correcting father of himself without contacting a profile manager as this is an obvious error....

• Change history Belt-467 14 may...

Maybe we should add a section on the error pages created (think just some people can update those pages....)

Suggestion add a section to the help pages...

Type of error: Obvious genealogy error please correct it if you have sources that 

--------

Type of error: Double check with sources If you are unsure please get in touch with the profile manager

Let me know if we should add something like that and as it is repeating text maybe have templates for type or error and recommended action ==> easier to update later at all pages....

{{Db_Errors_Type_Genealogy}} 
{{Db_Errors_Type_Contact}} 
{{Db_Errors_Type_AlwaysContactActiveManager}} 
......

---

I can add you and anyone else who wants to write documentation of errors to trusted list of DBE_xxx pages. Jan Terink set them as public instead of open. I don't know exactly why. Such templates would be useful to classify types of error and steps to correct. 

---

Please give me a name standard and names and i can create them....

My experience trying to create 2500 parish pages is that standard text and external links should be done using templates plus that you also get consistent layout.... maybe we would like to have icons in the future or G2G tags per error....;-) 

Templates created for Swedish parish project

---

@Aleš,

I set privacy levels to "public" to enforce (for the time being) a standard set up and layout. Have just now created a skeleton page that can be used as starting point when adding new error doc pages. Have set privacy level of existing pages to "open".

As apparently Magnus is going to develop some tamplates I will stop adding error descriptions for now.

---

You can continue with page creation. If something will change, it is easy to correct existing ones.

I already added links to 100 errors on error lists and on tree and location queries. So let's finish what we started.

Answer 6

I also agree with Vic,  there was a reason behind the Pre 1500 certifications and anyone, no matter what project, they should qualify for the certification same as anyone else.

Also agree that if you change someone's profile not only should you leave a comment, but list the source you are using for the change.  I am more for collaboration with PM's before changes are made.

PPP profiles I feel people should list what they feel needs to be changed along with the source backing the change, then let the project and PM's verify information.  Once verified then the project/PM's should make the change.  This is to ensure that there are fewer errors made on these profiles.

Answer 7

If you're interested in this topic and want a productive way to express your opinions, please see this question: https://www.wikitree.com/g2g/253636

Thanks!

Answer 8

OK, there is a problem coming up with so-called 'errors', as follows:

A name such as Beaufort dit Brunelle, the last name at birth box gets the Beaufort, and dit Brunelle goes in other last name box.  BUT:  ''dit'' is actually not part of the name as such, so has been put in like this:  (dit) Brunelle.  Seems the DB project is throwing this up as an error.  It shouldn't.

dit is not a name, it's closely akin to ''aka'', although not an exact translation.  Comes out more like Beaufort called Brunelle if we translate.  We find the terminology used in records, but in addressing the person directly, the dit would not have been said.  

 

somebody please fix the DB errors program on this.

Comments

I also see a lot of errors for French names like (dit) Laprise or (dit) Lacoursière where the error report says "Separators in Current Last Name" meaning the (dit)... 

I found not all profile creators want you to remove the (dit) and make it dit Laprise or dit  Lacoursière. 

Does that have to be an error .. because it makes the error report longer with more corrections to names that may be just cosmetic in nature. It takes alot of time away from correcting other errors just to remove the (( )) around the dit or dite etc. 

---

As I understand dit is not a name. It means aka. and Aka is automatically added in front of LastNaneOther. 

Random example:

https://www.wikitree.com/wiki/Desrosiers-306

Louis Narcisse (Narcisse) Desrosiers aka (dit) Lafrenière

here dit is not really needed. Or is it?

---

I am not sure Ales,, I saw a discussion on g2g here: https://www.wikitree.com/g2g/322655/why-is-dit-not-an-alternative-to-aka?show=322655#q322655

And in working with the errors report I find several names with that on it but because of the error on the left hand margin saying "separators", I always assumed they want the separators off, not the "dit" or "dite" off.

And that is whether the name is in other last names or part of the current last name.

---

We are dealing with French names here, so aka is incorrect.  The question has previously been asked whether aka could be replaced by dit for French names in the program.  Wikitree guidelines do state we should use the language of the person and place for names.  So you would not take the name Brunelle and change it to Brown.  Dorothy has included a link to one such discussion.

I suggest that the program be amended to ignore such separators, plain and simple, as well as their contents.  So in search function, it would only read the Brunelle part in ''(dit) Brunelle.''

Ales, your own name is a case in point, you need the keyboard that gives that particular sign over the s.  We make do calling you an incorrect name because we have no choice here, but the DB error pjt should not throw your name correctly written up as an error.  Same thing applies.  AKA is English, not French.

---

So from your answers I conclude that dit is aka in french.

And you can not do this in such way. Will you also add french translation for birth location like 'Lieu de naissance Paris, France' and similar to other fields?

Aka is added by wikitree to page contents automatically. If and when Wikitree will support localisation, this word will be translated and then you will have 

Louis Narcisse (Narcisse) Desrosiers dit (dit) Lafrenière

I will talk to Chris about this.

---

Which points up the first problem straight away.  Do you localize for the user or the ancestor?  If a French person works on a German ancestor, which bits are in French and which bits are in German?

---

Localisation is done for the user. Since I don't speak french, I don't know what dit means. Nor "Lieu de naissance". For me it should be "oz." and "Mesto rojstva" in slovenian.

---

 Aleš Trtnik 

 to help  .. this was a good discussion on the subject .. good for review .. https://www.wikitree.com/g2g/322655/why-is-dit-not-an-alternative-to-aka?show=322655#q322655

Dorothy Barry and Danielle Liard  are speaking to some of the points previously made in the referenced discussion .. 

C'est Bon .. 

---

I read whole discussion, and I still don't see the difference between AKA and DIT. It looks like the same thing in different language.

But i can add (dit) and (dite) as an exception to 791 error.

Answer 9

_{​I thought we get error list to let us know we have mistakes on our trees? If we do not fix them, the errors stay on the list. Isn't this issue why anything changed can be undone? In error no matter what?}

Comments

Cynthia, the definition of what is an error is what is at fault here.

Answer 10

ok, these errors codes are becoming outrageous, getting a 554 error because a dob is approximate year only, in older profiles, this is often the case.  I think this DB errors program is going too far, so often it throws up unusual names as possible errors, please, can we cut it back to actual obvious errors  like child born after death of mother or such like?

 

There is also the error coming out of multiple first names, which is being addressed by putting hyphens between the names.  Like Marie Madeleine is what is on the record, being converted to Marie-Madeleine.  That is in fact an error as such since hyphens were NOT used in earlier records.  My own baptism shows Marie Denise Danielle, with Marie being a generic given to all girls in a certain time period, Denise being my godmother's given name, and Danielle being my own given name.  They all figure in the record, but Marie and Denise are NOT middle names either.  We need to resolve this program and the various ''errors'' it is throwing up, since some of them are not in fact errors.

Comments

>> I think this DB errors program is going too far

See it as a help to do better genealogy. You will never get everything ok. 

I would like to see the database error project help me "rank" profiles so that I find easy that this profile has less sources than other profile form this country and time period 

 

---

As described on help page 554 Wikidata - Imprecise birth date on wikipedia they have precise date, so it is easy to check there, find source and enter precise date in profile and of course add the source. 

Unusual names are very possible typos. Error 511 is exactly that. It tells you that the name is unique (appears less than 5 times). You have quite a few of them http://wikitree.sdms.si/function/WTWebUser/errors.htm?UserID=6690041 , since you are using Hyphens. 

Why are you using them? I checked Liard-40 with name Marie-Adélaide It appears 4 times on Wikitree. If you check Marie Adélaide it is used over 50 times, so most people are not using hyphens.

 

---

I think I in the Birth cert had Carl-Magnus as my name but I never use the dash 

Maybe for names with dash also check it without 
 

• Marie-Adélaide ==>
  • Marie-Adélaide
    • Marie
    • Adélaide

If less than 5 on also Maria or Adélaide then flag as possible error