db errors on non-open profiles - more support needed for monitoring

Question

When trying to fix errors not on one's "own" subtree but more globally, i.e. by error code or by location the significant amount of non-open profiles are a problem.

Posting a comment on such profiles, stating the error(s) and asking for correction currently is the only available option if one wants to do at least something.

It would be nice if we could record that a comment was posted, in the same way as currently supported for reporting false errors. That way we avoid processing such profiles again next time. Also a better view of the progress in the project is available then..

Of course the best solution for non-open profiles would be to automatically post a comment. But that would require bot-like functionality.

Comments

A small step is that if you use the new template db_errors it's usage can be tracked 

---

@Magnus

Nice template!

Practical  issues:

• Non-open profiles cannot be edited, so can't add template to bio
• Can add template in comment (only format 4, else message "No WikiTree ID")
• Just template in comment probably not clear enough for PM. Need surrounding text asking for action
• Template in comment is not tracked
• Tracking is not useful for individual "error fixer"

 

---

Maybe you or Aleš Trtnik has some suggestion on a standard text...

The good thing with a template is that its easy to update the text later....

Aleš Trtnik spoke about that we should maybe have some help text/video......

------------

The template without parameter WikitreeID use a variabel {{FULLPAGENAMEE}} that as you see is not implemented on the comment page by Chris..... maybe Chris has a better variable/solution or we have to use 

{{Db_errors|10|Terink-1}}

in the comment field on protected profiles. The track function is standard Wiki inbound links I guess that the comment function is created specific for Wikitree and doesn't support all standard functions like variabes and inbound links...... 
 

---

{{db_errors|10|Sälgö-2|Y}} 

Parameter 3 creates some more text.....

---

@Magnus

A general text could  be like this:

---

This profile has been identified to have problem(s), see

Problem report

Please correct, if possible.

For more information please see Database errors project or ask a question at G2G with tag db_errors.

Thanks,

<ID of submitter>

---

Best text of course would be a "how to fix", but as the template will have no knowledge of the actual error(s) that won't be possible.

---

1. Should we always have this text or a parameter for getting this text?!?!?
2. As the template can be used also on a users profiles who has no errors maybe say something like below...
   
   Wikitree has started a database error project with the intention to correct data in Wikitree. Please see generated report on this profile
   
   Sälgö-1 report
   
   For more information please see Database errors project or ask a question at G2G.
   
   And please correct errors, if possible.
   
   Thanks,
   <ID of submitter>

---

@Magnus

1. Yes, I think  a parameter would be nice, because:

2. This parameter could control the actual text produced, for instance

C(heck) : for use in bio, produces your text

P(roblem) : for use in comment when error(s) are present, produces my text

---

Please give me the C and P text you would like..... maybe C could be default.....

---

@Magnus

"C" as default Ok.

Maybe allow thrid parameter to be name of a freespace page where the text to be poduced is stored? Would allow for customized text in any language. For example parameter "Space:Terink-1_error_text_305" could have the Dutch text for how to fix error 305.

---

"C"-text:

Wikitree has started a database error project with the intention to correct data in Wikitree. Please see generated report on this profile

Sälgö-1 report

For more information please see Database errors project or ask a question at G2G.

And please correct errors, if possible.

Thanks,
<ID of submitter>

---

"P"-text:

This profile has been identified to have problem(s), see

Problem report

Please correct, if possible.

For more information please see Database errors project or ask a question at G2G with tag db_errors.

Thanks,

<ID of submitter>

---

If you put Generation=0 user will get the errors only for that profile, which should be default in comments.

About Description of error, if we make a freespace page for each error with description or video of what might be a problem and how to fix, you could include link to that page maybe with third parameter E106. I would also put link to such page with each error, so users could understand the problem behind the error. 

For now the only description is on Project page: http://www.wikitree.com/wiki/Project:Database_Errors#100_Person, but it is only some basic info, that I add.

 

---

Or link from the report to how to fix the error

---

I wrote that. Maybe it wasn't clear enough. We just need someone to write it. I don't have time for that. But I will explain technically if there are some uncertainties.

I would also put link to such page with each error, so users could understand the problem behind the error. 

---

I will try to write the error descriptions and how to fix, assuming:

• Separate page for every error code
• Page is freespace type
• Pagename: Space:Database_Errors_Project_Error_Code_nnn where nnn is numeric error code.

Are these assumptions ok?

Privacy level of these pages?

---

I should use shorter names with a prefix e.g. DBE_202 and also have some common header that is included on all error pages ==> you change just in the common header and it is displayed on all pages. Could be useful to announce changes on all pages, new help info,videos, statistics.... 

I created a structure for Swedish Parishes then I had one header page
Space:SPC_Header containing a infobox that will be presented on all pages ==> could contain link to DBError project page etc,,,

For every county in Sweden (24) I created Space pages that included Space:SPC_Header e.g. 
Space:SPC-AB
Space:SPC-C
SPC-D

The parish pages 2500 will just contain the county page it belongs to eg.
Kila (U) --> contains SPC-U ---> contains SPC_Header 

Below a picture of Category:Frostviken_(Z) that contains Space:SPC-Z that contains SPC_Header in the picture below you can see what parts is included from the header page and the county SPC-Z page ==> easier to update.... 

Example: Category:Frostviken_(Z)

---

Jan, I agree with Magnus on shorter page name. I will add this link to each error, that means a few million times, so let's make it short. Space:DBE_202

I have nothing against include page. It can always be useful. 

You can take my descriptions on project page and continue from there.

Should we have a few subsections like

=== Description ===

=== Possible causes ===

=== How to fix ===

=== Technical ===

I suggest you write one page, and then we discuss about contents, layout, ... 

---

Shoot!

Note:

• No header (as suggested by Magnus)  implemented yet
• In "checks" section I assumed the date qualifier "after this date" is not part of your checking. Correct?

---

Caption can be longer. Just page name should be short.

I am missing link to project. It will probably be in include file.

Action section should have another solution: If birthdate really cannot be established, delete it.

I don't quite get "checks" section. I use the date of database dump.

---

No include file (header) yet, concentrating on the error first. Better like this?

---

I would put also text of the error in the page title. Maybe like this

Error 101 Birth in future (Database Errors Project)

In Checks performed section also mention private profiles with date written like 2020s

---

I added to Space:DBE_101 as a test

{{Space:DBE_Header}}

• has a sample banner
• a category DB_Toolbox ==> gather all stuff we create at one place

{{Space:DBE_Footer}}

• has just a footer with project related links.....

My experience is its good to have them even if they are empty....
Feel free to do what you want with them...
 

---

@Aleš

In Checks performed section also mention private profiles with date written like 2020s

I don't know what check this is, so I can't create text. Please explain, and I will add text.

---

@Magnus,

Ok, thanks for supplying Header & Footer

---

It is for Error 101. Private profiles have birth and death date written as 1950s to indicate a decade. So 2017-01-01 is written as 2010s and I don't mark this as an error. But 2020s I can say for sure, that date is in the future.

---

If this setup is OK with you I can go on and do all error codes like that.

Note that I now name the pages: Data Error nnn: error text

I think to the user it is irrelevant that the data is stored in a database and also that the error was reported as part of a project. The bottom line is that there is invalid data, and that's what the pagename now says.

 

---

The page looks ok to me. As you will make them, I will review them and add links to them also on error reports.

---

Ok.

A question about dates: do you get the qualifiers (before this date, after this date) in the database dump. If so are they in any way used in the error checking?

---

I do have this qualifiers, but they are not used in error checking for now. When someone will complain, i will probably add them. Also if month or day are missing, I consider them as 1.1. I will correct this in the future.

If you need any explanation for help pages, write to my email, as I will see that sooner and you will get faster response.

Answer 1

Communication in this thread now focuses on the new and nice db_errors template. Therefore I am afraid the most important part of my question gets no attention:

It would be nice if we could record that a comment was posted, in the same way as currently supported for reporting false errors. That way we avoid processing such profiles again next time. Also a better view of the progress in the project is available then..

Comments

I lost track on too many G2G threads.

Here is my discussion I had with Carol over PM. It might be of interest here.

>> Hi Aleš,
>> You are doing such brilliant work here on wikitree, and inspiring a lot of us to help you with it.
>> I'm currently working my way through the duplicates that you have identified under error code 106, for the years 1900-1999. Have raised 67 merges so far, and covered just a small fraction of the errors, so it's going to be slow going. 
>> This is just a suggestion that could go on your 'to do' list and that would help us with the corrections. When you run another DB dump and produce another error list, could the duplicates that have a 'pending merge' be excluded from the list. As I do each merge, I check for duplicates in parents, siblings and spouses etc, so they are all raised at the same time. I'm pretty certain that a lot of these merges are not going to be approved by their profile managers, so they will have to wait for the default approval before they can be actioned, and they will continue to be shown on the error report. It would be far easier to work from the report if it only contained duplicates that needed merges raised. 
>> Don't worry if this is not possible, but thought I'd write and suggest it, kind regards, Carol
>> This message is from Carol Keeling: http://www.WikiTree.com/wiki/Winton-239
 

> On 16/05/2016 17:00, Aleš Trtnik wrote:
> Pending merges are not part of database dump, so I cannot remove those merges. Downloading page for each user would be very slow so that cannot be a solution. I could add link to hide specific error for xx days (similar to false positive). How many days does default approval takes?

> 2016-05-16 21:24 GMT+02:00 Carol 

> Thanks for the reply, Aleš, that's a pity that you don't have access to pending merges, I thought it was a good idea anyway, and worth asking.

> Profile merges (I think) take a calendar month, so one raised today should get default approval on 16th June. Don't worry about putting the hide link in, I'm assuming that the next error list for 106 will come out in the same order, so I just need to start on the next profile that needs looking at. I'm keeping a spreadsheet of all the merges that I've raised and all the trusted list requests so I can keep track of where I am. 

> Kind regards, Carol

So I could hide the error for a month, to give user some time to repair it (holidays,...) and after that time it would reappear so different actions could be taken.

---

Aleš,

Yes, being able to hide the error for a month would be of help, so please implement this function!

Thanks

---

How many days is timeout period for merge?

---

Aleš,

 

I replied from the context of "non-open" profiels, not that of pending merges. Carol answered Don't worry about putting the hide link in so I guess she does not really want it.

Anyway, the help text on merging says:

If a Pending Merge is ignored for over 30 days it will be cleared for any Wiki Genealogist (any signer of the Honor Code) to complete, reject, match, or remove it as they deem appropriate. 

So I would say hiding for 1 month should do, both for "non-open" profiles and "pending merges".

---

I support the "hiding link" for 30+1 days.

---

The idea of putting a hide link in sounded like an awful lot of work just for me, which is why I said not to worry about it. I would certainly use it if it were implemented. I assume that once I'd initiated the merge, then I'd click on a 'hide' symbol for the error, and it would disappear from the list, then magically reappear after about 30 days if I hadn't actioned the merge by then. And it would also be possible to monitor how quickly my error list was reducing, which is another bonus that I hadn't considered.

And overnight it's now all been implemented, that was so quick!, thank you, will start to use it immediately

---

Done. See http://www.wikitree.com/wiki/Space:Database_Errors_Project_2016-05-15#Temporary_hidden_errors