DB_Errors report 2.0 tighter integrated in Wikitree?!?!

+25 votes
615 views

I am a big fan of the Project Database Error and feels it change focus working in Wikitree to better quality.... 

My suggestion why not integrate it even tighter with WikiTree. That we get a section like below on all profiles with error reported on that profile? The Menu option and watchlist integration is excellent but not user friendly to find...  

A World Wide Family Tree must have quality focus to be a success and I feel Project:Database_Errors is a great step in that direction

Sample profile.....

WikiTree profile Williams-2657

Dont know if it possible and if Team Chris and Aleš has time/resources....

Technical it feels like open up for a <iframe> section</xxx> which is not possible with a template so it must be done in the WikiTree code..... plus have a new small report.... based on WikiTree-ID generated at http://www.sdms.si:92/function/WTWeb/

 

WikiTree profile: Richard Williams
in Genealogy Help by Living Sälgö G2G6 Pilot (312k points)
retagged by Dorothy Barry
The danger is that people will just do arbitrary fixes to make the errors go away.  You could reconnect the father, but the whole thing would still look very dubious.

I wouldn't touch it without sources, but I think we've got two Richards in one here.

Sure RJ Horace

What is the alternative? Hide and seek ;-) 

I think knowing that we have a potential problem and inform people reading it is a way of showing that you care and are serious....

Not showing known problems and people start duplicates a WikiTree profile like above has no winner is my feeling

 

But what should we do with the vast numbers of known problems that the system won't find?  The impossibilities aren't the only errors.

We're in danger of giving out the message that anything goes so long as it's within the bounds of biological possibility.

But what should we do with the vast numbers of known problems that the system won't find?  The impossibilities aren't the only errors.

No but is that an excuse to have a system with 47 673 profiles with error 205 Father is too young or not born

We're in danger of giving out the message that anything goes so long as it's within the bounds of biological possibility.

I don't follow you. Before you got the database error report what message did Wikitree tell?? Upload what ever you want and we trust you as we don't check anything....

If you should add quality you need to focus on that. 

What to do?

  1. Use what we have and that is the Database Error project and try to make it easy for people to understand that it exist ( a new menu option is good but we have 70 menu options today in WikiTree...
     
  2. Start define new errors
     
  3. Start define standards what should be in different fields and define new errors to fnid those problems
     
  4. If Wikitree people are interested and we get some code guru like Aleš Trtnik we can add dimensions if a profile is sourced or unsourced..... add more logical tests..... I see a lot of "Ancestry imported" Swedish profiles were there is no logic how people travel in Sweden in 1800 ==> indication that it's an unsourced profile with no sources.... 
     
  5. We could start adding peer review and mark profiles reviewed and has passed some quality level etc...

RJ

I have not seen one positive comment from you on this project. Are you saying we should abandon it? Do you have any suggestion on how to improve quality?

Of course some people may make arbitrary, invalid fixes. But there are also many members very grateful for having been shown errors in their profiles, and eager to correct them properly.

One idea I have for new error (903?) is report profiles that have the WikiTree statement about this biography was auto-generated.  That usually means it's a gedcom dump, with no additional editing.  I might not try to edit these from the general error report (there will be thousands) but I would use it for my watchlist and for branches I'm related to.
Biography is not part of database dump, so I can not do this for now.
It's reverse psychology.  Support from me = kiss of death.

Having banged on for 2 years about the junk mountain and the obstacles to fixing it, obviously I'm all in favour.

And in general terms I'm especially in favour of sanity checks on profiles.  The bizarre thing is that we can't add an {{Unsourced}} or a {{citation needed}} to a profile without giving grounds for complaint.

I did a few dozen corrections, but I kept running into the obstacles.  How can WikiTree be helped when it's so paranoid and restricts access to the facilities?

It should be obvious that if people are allowed to get on with the job as quickly and efficiently as possible, the profiles that get fixed will massively outnumber the ones that get screwed up.

Actually this might change now.  I think the Corrections Officers will keep coming back for more badge powers and exemptions, and then they at least will be able to make progress.

They'll need to, because I doubt if more than a few percent of corrections will be done by PMs.  There's enthusiasm on G2G, but G2G readers are highly untypical WikiTree users by definition.

I do worry that the db_errors stuff isn't ready for all the users who haven't discovered the Edit tab yet.  They especially won't get why the box doesn't disappear immediately when they fix the error.

>> I do worry that the db_errors stuff isn't ready for all the users who haven't discovered the Edit tab 

As always with development its difficult to understand the direction of something i.e. db_errors I understood was something good but now I feel its much much more....

If people has learned to use internet then Edit is not a big step this is not rocket science.... and is ok to do mistakes...

If something is difficult maybe we find a solution for that....

But templates in biographies could be (and should be) put into the database dump.

4 Answers

+5 votes
Nice idea. We may want to wait until we have cleared out the lowing hanging changes. There will still be massive DB errors left after we cleanup the simple ones due to protected profile born in the last 200 years. Sending messages only gets your so far. in these cases. Adding something to the profile would be helpful, You may want to add a qualifier in the based openness of the profile. Perhaps encouraging the PM to open the profile, particularly if deceased.

Some PMs will poorer quality data will be in for a rude awakening when many of their profiles are tagged like this.
by Marty Acks G2G6 Pilot (165k points)

Maybe an option on our login Wikitree profile Special:Settings to turn on/off display Database Errors report ;-) - Default on ;-)

+7 votes
Aleš and I have been talking a little by e-mail.

He is willing to share his database table. It would only require a few fields. I think the error code and one or two person IDs. We could put this database on our server.

The biggest issue in my mind is the burden of loading another table every time a person profile is viewed. This would be a relatively small table but it all adds up. Milliseconds matter. It's not just page responsiveness for users. Google will rank us lower if our pages are slower.

We could mitigate this by only having the errors visible to logged-in members. Maybe we'd want to do this anyway. Or maybe we'd want to highlight the errors for everyone, like how Wikipedia flags potential issues so the reader sees them. ???
by Chris Whitten G2G Astronaut (1.6m points)
I think you should show only errors of the profiles in question. At the moment we have a little over 1M errors, so every 10th profile would have 1 error added.

And you don't need to include information about the profile, since it is already displayed on the page. Also related profile has a link on the page (parent child, spouse, sibling). So it would be very little additions to the page.

This seems like a good idea, at least for the profiles displayed to logged-in members.

However, before this is done it would be good to find a way to focus in on the specific profile(s) most likely to be the site of an error that affects multiple profiles. Many of the errors that I currently see in my Error Report are situations where a single error on one profile adversely affects a bunch of other profiles; I don't see it as real helpful to flag all of the affected profiles as having errors.

Example (thinking of one batch of errors that Robert Keniston fixed a little while ago by an edit to one profile):

If a woman's death date was set to equal her date of marriage (apparently because someone was ultraconservative in creating dates in their gedcom, and the import utility picked the earliest date in the range of "1640 to 1700"), yet she married and had 12 children, the profiles for her and all 12 children would show errors in the database. I don't think it would be helpful to put an error message on all 12 profiles. Is there a way to search for the likely source of the error (in this case, the woman's profile) and add the error message to that profile only?

It is an easy task for human to see the problem is the mother, but for computer it is much harder to decide such things. All possible combinations should be taken into account and that is too much work for a little effect.
putting the database on the WikiTree server is both good and bad for other reasons...

I can't access the current errors site from work due to firewall restrictions. (maybe that's a good thing)
+4 votes
The Richard Williams situation might be messier than I thought.

The Arborists have a scary fluorescent template for Bad merges.  Perhaps the Corrections Officers could have one for the Serious_Data_Problems category, to put on any tangles they don't fancy tackling straight away.

As with the bad merges, this might warn people off making things worse.
by Living Horace G2G6 Pilot (668k points)

Yes some kind of Red Alert flag would be good

 

I miss that we don't quality assure profiles more

  1. I would like to see Featured profiles that should be examples of what can be done
  2. Profiles that has been reviewed to follow the GPS - Genealogical Proof Standard or some kind of other reviews see G2G 
  3. or warn
    1. This profile is from a Gedcom import with 80% unsourced profiles
+1 vote
That sample is a good one. One is the man is Father of child by the name of John Williams b 1648. This is a false error only if he had 2nd marriage prior to the one in 1671.
 2. Margaret Cook may very well be the same person twice, but this indicates that they are not merged. Marrying same day says probably same person.

This too is not a false error until the 2 Cook profiles are merged.

Generating a single error report would not take these things into account since they are not part of the profile. The error of "Unusual name spelling" could be removed.

Perhaps a report could be made with "Gender Missing" and allow for faster correction.
by Jon Czarowitz G2G6 Mach 4 (47.1k points)

Related questions

+6 votes
1 answer
+4 votes
2 answers
+7 votes
0 answers
+7 votes
1 answer
+18 votes
1 answer
+11 votes
0 answers
+132 votes
94 answers
+3 votes
2 answers
107 views asked Mar 8, 2014 in Genealogy Help by Nick Apodaca G2G Rookie (220 points)

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

disclaimer - terms - copyright

...