Wikitree requires that we update WikiData?

Question

I just had my first DBE 544 error.   the help page for that particular error indicates if the profile on WikiTree is the same person as the profile on WikiData, I should do the following.   This is the case that I am referring to....it is not a false error, I am not going to ignore it, I try to get my suggestions at "0" as any good WikiTreer should.

Profile on WikiTree represents the same person as profile on WikiData. Add statement in Identifiers section for WikiTree person ID and enter mother's WikiTreeID.

That means that I know have to learn how to update WikiData to clear an error?????

Comments

I hope Aleš will chime in; I've been under the impression that there's an automated process.

Indeed, I'm almost sure there is ... I've seen at least two instances of an incorrect link being repeatedly re-added to Wikidata after it was removed several times by Data Doctors who master those errors. The WikiTree profile and the Wikidata item did not represent the same person at all, but the link from Wikidata to the WikiTree profile was repeatedly added back.

---

I want to make sure all of you answering understand that I am quoting from the help page for that suggestion.   It says to fix the WikiData page.....

---

didn't realize that...

... that does seem an odd thing to suggest, doesn't it?

---

Actually, fixing Wikidata is the only way to clear the suggestion.

What some answers say is that it is just not necessary to clear it. It is not an error, it is just supposed to be a hint.

Problem is, it does surface on our report like an error we "have to " correct.

---

So, if I read through what Ales said below....I do not need to do anything if they represent the same person, and the link on Wikidata needs to be updated, because he has a process to do that....

---

Yes, but if the suggestion doesn't go away in a week or two, it probably isn't find as very likely to match and also probably won't in the future. In such case manual action is in order.

Answer 1

That means that I know have to learn how to update WikiData to clear an error?????

Not necessarily. I think an important point to remember is that these are Suggestions, and not everything you see in these suggestions will be errors (like when you accidentally transpose a date and make someone have kids 40 years after they died).

There are groups of people that will have a more specialized area of interest in the Suggestions lists, and if help is needed, you can always ask in G2G . I am comfortable with WikiData, but when I looked at the profile I think is in question here, I stopped based on the differing information (names, dates, etc.), so I am afraid I will be of no use unless you can verify the profiles are indeed a match.

Answer 2

As Steve says, DBE 544 is only a suggestion. I find it and other 5xx Wikidata "errors" very useful for people who care about the Wikidata-WikiTree sync (I'm among those). 

BUT calling such mismatches "Data Error" is certainly too much. Those things are not "errors" like e.g. a birth date posterior to marriage date, whatever.

And if you don't care about Wikidata, just ignore those. 

Comments

Above comment actually mine - login error 

Answer 3

WikiData is an external website. And they, like us, are not immune to errors.

Remember, the Suggestions report is not necessarily saying we're wrong, or even who's wrong, but merely that there is a discrepancy.

Just like Find-a-Grave errors, if you have access, correct their errors or don't. But if our data is correct, then mark it as a false error.

Nobody is saying we have to correct their errors.

Answer 4

I worked diligently for several weeks to clear all of the DBEs for myself and the Magna Carta Project and then stayed on top of the error reports each week. But I don't work in Wikipedia (or FindAGrave) - just WikiTree - so I have a growing list of DBEs that won't be cleared. I had been marking them as false after verifying that the relationship in WikiTree was correct and the wikidata had nothing useful to add, but I'm told that's not the correct thing to do.

Comments

I have been trying to stay on top of my suggestion list, too, but now on top of the errors I get for the bunch of profiles not connected to the World Tree, I have 68 "Clue for father" and 28 "Clue for mother" suggestions. Not the same error as mentioned in this thread (I handled those) but it's telling me that a profile has a parent in WikiData and there is none on WikiTree. I try my best to ignore those since a) adding the parents in question is not my priority and b) if I add them it will ask me to the next generation the next week, etc.

If these "clues" could be in a separate list and not bug us as "errors", it might help.

---

If there is something unclear about this new hints or Clues, ask with an example. I will help you understand them. This clues were created after connectathon and can be used to add missing relatives to the profiles.

---

If you do add the "missing" parents, you'll then get the 544 telling you to make the connection at the Wikidata end.

---

Correct, RJ, and I will also get "clue for mother" and "clue for father" for their parents and I'm back to square 1, only x2. I checked and Wikidata really have lots of entries. Most of them with no info and just a name entered in Cyrillic.

I finally got used to them, I just ignore them.

---

If the data on both ends matches enough, the connection is added automatically by me. I add several hundred of them each week. Name must be partially similar and the date must approximately match. So in most cases there will be no 544 suggestion.

---

Oh, so there is an automated process. I thought there would be.

What can we do if that automatic addition adds an incorrect connection? I've seen that happen sometimes, I think it was when a profile was confused with a sibling with a Wikidata entry with but no WikiTree profile. The connection had been removed several times but it kept coming back. Is there a way to prevent this? I managed to resolve those I found by creating a profile really matching the Wikidata entry, but there will be situations, where this is not possible. Wikidata has errors.

---

Note  @Aleš if I see that happening again I will make sure to send you the example.

---

Isabelle, I also noticed those Cyrillic entries without the associated wikipedia page. I added them as possible since I can read some russian. They are not of much use at the moment, but the things can change in the future as more different sources are connected together.

For instance last week a lot of data was imported for european aristocracy and I added 2000 new connections to WikiTree and those profiles are now checked for consistency. I expect such imports will continue to happen and overall data quality will improve.

---

Yes there is a semi automated process. I can provide more info on it.

Each week after the dump import, I also import WikiData entries. Then I first check for missing WikiTreeIDs that can't be connected. There is 3-4.000 of them. Most of them are to Unlisted profiles. I don't delete them, since GDPR things might change in the future. I do check the 50 entries with death date entered and correct most of them. They are usually a result of a merge on WikiTree. in some cases a person recently died and I enter a death date from WikiData and open a profile. Once a month or two I also compare them against unlisted profiles and correct other merges, that don't have a death date. This is the first step.

During the dump import, all references to wikipedia and wikidata are extracted from the profile and checked against WikiData entry if they represent a person. There was 159569 such links and 109701 were to a person last week.

This profiles are compared against wikidata and if names and the dates approximately match, I add the connection.

Then I reimport WikiData and compare the relatives on wikidata against the relatives on wikirtee and add the matching ones. I repeat this step once or twice depending on the number of added connection.

I also add connection based on FindAGrave Property on WikiData if they match our FindAGrave link.

All this lists are created by a query and I visually compare the data and decide to connect a profile or not.

The whole process takes me an hour each week. Occasionally I actually inspect the data and profiles on questionable matches and decide for each one if it is a match.

Answer 5

Stop calling them ERRORs. I did 2 years ago and I almost always refer to them as SUGGESTIONs. It was corrected on most places on wikirtee and wikitree+.

Each suggestion can be one of 3 types (Error, Warning or hint and the type is defined on the help page). It is a Hint in our case.

In this case it is a hint of existing mother on WikiData and WikiTree. If they are actually representing the same person, they should be connected so other checks can be performed. If they aren't, then there is an error on one site or the other. I have seen the errors on both ends. If someone is not familiar with wikidata, ask for help in G2G and provide some info, if you are familiar with the profile. Many people know how to use also WikiData and will help.

Comments

Ales,

The title of the page linked in the question is "Data Error 544 Wikidata - Possible Mother on Wikidata".

The first line on the page is "Description of Hint - 544: Wikidata - Possible Mother on Wikidata".

The first paragraph includes a sentence that says "This suggestion indicates …", followed by a sentence that says "There is also a possibility of some error …".

Farther down, under a subheading "Action Steps", the paragraph starts "There are 2 possibilities of an error."

Below that is a subheading "Hint Status" that says "For each Hint you can …" but the next paragraph says "If there are related errors …".  The section ends with a link to "Status Help Page", but that page is titled "Error status Help Page" and uses the word "error" several times, but the word "hint" does not appear anywhere on that page, nor is the word "suggestion" anywhere on that page.

I am confused by words like "error", "hint", and "suggestion" seeming to be interchangeable, without any clear distinction between them.  I think it would help greatly if there were a clear indication of the difference between the meanings of these three words, as they are used on WikiTree, as well as consistency in how the specific items are identified, both in the help pages and in discussions about them.

---

Page Title is indeed incorrect. As you can see on 

https://www.wikitree.com/index.php?title=Category:DD_Suggestions_Help&from=DBE_793

it is quite inconsistent. Although newer pages don't use error any more. I will ask Project coordinators to unify the titles.

All others references are mostly correct. 

Suggestion is a general name for all suggestions. and it can be one of 3 types (Error, warning or Hint) this 3 types are also colored red yellow or blue on WT+. Recently The Clues appeared, that are even less important than hints but are in that group for now.

---

If language is not correct, then what is said is not what is meant; if what is said is not what is meant, then what must be done remains undone; if this remains undone, morals and art will deteriorate; if justice goes astray, the people will stand about in helpless confusion. Hence there must be no arbitrariness in what is said. This matters above everything.

Confucius

---

And have mercy on the poor non-native english speaker 

---

Bernard, is that a quote from the great Chinese philosopher named ConfuseUs?

… oops, did I mis-spell his name?

---

孔夫子 can sometimes be Kǒng Fūzǐ -ing indeed, but most of the time he is just boringly clear and to the point.

The really confusing one (and hence my favorite one) is 莊子, and for today's controversy I would particularly recommend this piece : https://ctext.org/zhuangzi/adjustment-of-controversies.

But, back to the former, I read here ...

 https://en.wikipedia.org/wiki/Confucius#Descendants 

... he has the largest recorded descendants tree, down to 83rd generation. Do we have at least a part of it here in WikiTree? I see on G2G archives a question about it from five years ago, without much follow-up it seems. https://www.wikitree.com/g2g/84522/can-wikitree-incorporate-the-kong-family-genealogy

Enough said. This is completely off-topic.

Answer 6

In my experience, WikiData can be an insidious source and purveyor of misinformation.

A couple of years ago, I dealt with a situation where notable "Y" had been misidentified in Wikipedia and FindAGrave (and possibly other places) as a man of the same name (call him "X") who was the father of another notable, "Z."

It took a very long time to get the computer algorithms at the heart of WikiData to stop insisting that "Y" was the father of "Z" (and WikiData kept updating the error, resulting in repeated instances of the DB suggestion here). I edited Z's profile in WikiTree, I created correct profiles for "Y" and "X" in WikiTree, I edited the English Wikipedia articles for "Y" and "Z" ("X" was an extremely wealthy man, but he apparently was very private and hasn't gotten a Wikpedia article), I contacted the FindAGrave memorial managers to request corrections (corrections were made). Finally,to prevent the errors from returning, I had to make edits in WikiData itself and some other obscure-to-me sources that WikiData apparently relied on.

The errors finally went away. I haven't checked these people in WikiData recently, but I'm almost afraid to do so because I remember how the errors kept returning after I had corrected them.

Comments

It's the Second Law.  In a system like that, the information will always degrade.  It'll have to keep being fed back in from an external reservoir, like books.

Only soon there won't be any books.  Then someday, somebody will think it's a good idea if the scans of old books can be "corrected" to say what they should have said.  Then there'll be no reliable information anywhere.

---

Yes, Ellen, a growing problem in Wikidata is the power of bots keeping data synchronized with a variety of external sources. Very handy, but if the said source is wrong, it's hopeless to try and correct "humanly" the data in Wikidata, you have to find out the faulty source and go there to ask for correction if you can.

That said, if Wikidata was always abiding by its own rules, every fact should be sourced. You can add two conflicting facts (e.g., different birth dates or locations) with for each one, the source, and one could be a WikiTree profile. But with no guarantee that the bots will not kill your conflicting assertions.

---

There are multiple cases where conflicting information is present on wikidata. You can have multiple dates for a person's birth date and If anyone of them matches wikitree, I don't report that as a suggestion.

I am even thinking of starting to populate other Wikidata information with WikiTree info and a reference to wikitree as a  source.