WeRelate talk:Variant names project

I've been away too long, and wonderful things have been happening! Glad I could finally make it back. But I'm trying to catch up. The variant names project sounds marvelous. This always has been a special interest of mine. But I need to make sure I understand what's happening in reviewing and editing here. In doing some browsing, there were additions to surnames that made absolutely no sense to me. One example: For the surname 'Mominie', I did not find Mominee, or Montmeny (the name it came from), but I did see that Monaghan and Minyan had been added, names that were of totally different cultural sources. I added the variants I knew, but I was leery of deleting the others. Were these name changes known by the person listing them? I thought I should ask some questions before I just dove in. If I'm repeating questions discussed elsewhere, please let me know, and just point me there, please. Like I said, I'm coming in late....

I'm figuring that we should be adding name variants and spelling variants that: A. We have seen as variants among people we have researched. B. We have seen as misspellings on censuses and official records, especially those that were likely phonetic or cultural and so are likely to have been repeated. and (possibly?) C. We have seen as plain old name changes, such as Stachowicz becoming Stack.

I'm figuring we should be removing name variants and spelling variants that we know, especially from previous Soundex and Metaphone searches, are *not* related surnames. For instance, 'Schreiber' has a soundex match with Sherbrooke, and 'Robinson' matches out with 'Rappahanock'.... Those should be deleted. Others are iffier. I have seen lots of variants of Hoover, but I have always found Hoppers and Huffs completely unrelated. Doesn't mean that others have the same experience. Ditto Goss and Gossett. And how are 'Pc' and 'Pk' variants of Peck?

I am more comfortable adding things I know of, and less comfortable deleting, I guess. Any guidelines? If some of this stuff is not deleted, we will have almost as massive and senseless output as Soundex...

Thanks, --LindaS 19:28, 30 January 2012 (EST)


Thank you for getting involved!!

I'd recommend being comfortable deleting. Most (90%+) names were added using a computer algorithm. It's better than Soundex, but it's still a computer. Also, I erred on the side of including extra variants because I figured it was easier for people to take variants away than to add them.

On the watercooler we're talking about possibly adding back a source for each variant, and the ability to watch specific names to be notified of changes.

On the names to add, I'd definitely add your A and B names (variants and misspellings). Regarding C, I'd say add only if you think multiple people would have changed their name this way. For example, I have an ancestor that changed their name from Husser to Halverson, but I wouldn't add Husser as a variant of Halverson because I believe changing from Husser to Halverson was done only by this one person.

Thanks again!--Dallan 10:44, 31 January 2012 (EST)


Thanks for the response, Dallan.

The 'add' parameters are what I would expect, and make a lot of sense. The 'delete' function is now a lot more comfortable, knowing how many of those added names got there. I'll stop by the watercooler and read through what's been happening, too.

--LindaS 18:50, 31 January 2012 (EST)

Topics


Unknown as a given name in searches [8 February 2012]

The system doesn't allow to search given name 'Unknown' so I thought I'd mention this: I hope that when someone searches for Unknown Smith, the results also return Son Smith, Daughter Smith, Mr. Smith, Miss Smith, Unnamed Smith and Stillborn Smith. These are all names I've seen in various charts entered in the given name space. Yes, these given names should be changed to Unknown during the upload process, but sometimes they don't get changed. --Janiejac 16:32, 7 February 2012 (EST)

The system does allow you to search Unknown now, as well as Son, Daughter, Mr, etc. It doesn't search them all together though; you'd have to search each one separately.--Dallan 20:20, 8 February 2012 (EST)




Why two checkboxes? [3 July 2014]

On the variant names search page, in the computer variants column, why are there two checkboxes beside each variant? Is one for Confirm and one for Remove? If so, which is which? --LulaBelle 00:03, 21 November 2012 (EST)

The one on the right is for remove; the one on the left is for confirm.--Dallan 00:53, 21 November 2012 (EST)

I made a mistake and Checked the wrong boxes for the Surname Gallahue. The names: galka,galle,gallee,gallice gallie, and gallou do not belong there, can you please remove them.

Thanks--Gallchú138 09:35, 2 July 2014 (UTC)

I just checked and it looks like they have already been removed.--Dallan 03:22, 4 July 2014 (UTC)

Evidence for surname relatedness versus similarity [27 June 2015]

I think this is a great resource for similarity between text strings. Do we have a resource or best practice for recording relatedness through known name changes and origins. A particular surname today might have descended from two or more origin names and just converged over time. For hard core geneaology, building a resource which could accommodate evidence-based relatedness would be good. --ceyockey 17:32, 26 June 2015 (UTC)

It seems like a good idea, but I'm not aware of any resource like this.--Dallan 05:51, 27 June 2015 (UTC)

Big mistake - please fix [23 November 2020]

I was looking at variant first names for Maud, and I ticked some in the first column instead of the second. This removed them instead of confirming them. This is disastrous. Please fix.--Edgarw 19:05, 23 November 2020 (UTC)

PS Is there a better place to post urgent requests?

PPS I now realize I can add them back by hand... :)

If you have urgent requests, the best thing to do is email me directly: dallan at werelate.org. The best thing to do is to add them back in by hand (sorry). But you can see what you removed by going to the log Special:NamesLog so it's not too bad.
It looks like you've done that already ;-)
Thank you for your work to improve the name variants!--Dallan 19:52, 23 November 2020 (UTC)

Adding comments for existing relationships [5 March 2022]

There are a lot of Confirmed Variant entries that do not have a comment associated that indicates what the basis of confirmation is. For instance, Surname fries with confirmed variant freese. Is there a way to add a comment to an existing relationship? I did not see one when looking through the interface. Thanks. --ceyockey 16:04, 5 March 2022 (UTC)


There isn't a way to add a comment to an existing relationship unfortunately.--Dallan 00:14, 6 March 2022 (UTC)


Working on a better way to generate computer-variants [5 March 2022]

I started this project many years ago, and the computer variants leave a lot to be desired (in my opinion). For the past six months I've been working with FamilySearch to develop an improved way to generate computer variants and to automatically-confirm many of the existing variants. The new effort is based upon a list of all (tree-name <-> attached-record-name) pairs in the FamilySearch tree and uses modern machine learning techniques to learn from those examples when generating name variants. So far the results are very promising. Sometime later this year I plan to combine this project and the FamilySearch project. If anyone is interested in more information on this effort, please let me know.--Dallan 00:24, 6 March 2022 (UTC)


Acommodating middle names? [7 April 2022]

Wondering about a name like "Ida May Ralston" vs. "Ida Mae Ralston". Should in this case we just treat first and middle as "conjoined" so that we would add "Ida May" with relatedness to "Ida Mae", which would render as "idamay" and "idamae"? Thanks. --ceyockey 23:11, 6 April 2022 (UTC)

Middle names are treated as separate names, so you'd enter Mae as a variant of May. (Te and Ten are treated specially, since they're considered to be prefixes, not separate names)--Dallan 01:31, 7 April 2022 (UTC)

Thanks, Dallan. --ceyockey 17:16, 7 April 2022 (UTC)


Not saving my input [13 June 2023]

Could the functoinality be down at https://www.werelate.org/wiki/Special:Names ? Today is not the first time I've detected a failure to save a variant with a comment. Today I was trying to document relationship between "yockey" and "yocky" based on a generational transition found in on of my family lines and saving. The save attempt led to url > https://www.werelate.org/wiki/Special:Names?name=yockey&type=s&updated=true , but the new value was not included in the variants list. The URL before save was https://www.werelate.org/wiki/Special:Names?type=s&name=yockey . Thanks for taking a look and confirming or refuting my experience. --ceyockey 17:40, 22 April 2022 (UTC)


This is not a persistent thing - works for me most of the time.

Problem with a specific variant. 'boyd' does not seem to like addition of 'bhoid' as a variant. Thoughts? --ceyockey 14:51, 4 November 2022 (UTC)

I went ahead and added bhold as a variant of boyd. I'm not sure what happened earlier. Please let me know if you run across other problems.--Dallan 00:35, 5 November 2022 (UTC)

I just tried to add 'jervoise' as an alternative for 'jervis' and the attempt failed - save unsuccessful with no error message. However, I note that 'jervoise' is present as a SOUNDEX variant of 'jervis'; could this be a reason that the save failed? --ceyockey 01:15, 6 November 2022 (UTC)


The interface is fully functional for me now. Regards --ceyockey 00:39, 30 April 2023 (UTC)

I spoke too soon. I am still clearly unable to use the add function under Confirmed Variants if I do not ALSO check one of the Computer Variants to be included in the variant addition activity. Latest fail - For the surname mgrath, tried to add mcgraw, which is not listed among the computer variants, alone via the Confirmed Variants addition field and the add failed to save. Regards --01:00, 1 May 2023 (UTC)

After adding the variant, you still have to click the "Save" button way down at the bottom of the screen. I just added mcgraw to mgrath and it worked. It's not a great User Interface, I know. I'm sorry for the trouble. I'll add a reminder that you have to click "Save" in the next day or two.--Dallan 04:08, 2 May 2023 (UTC)
I'm still having problems. I have tried successfully in the past marking a 'computer variant' as needing to be removed and this can let me add things - this worked for surname aalto, where it was not listed in computer variants and I had to add manually, and I removed elt to allow the save to go forward. However, I tried that trick again with abbondi and it did not work; I did a second edit attempt where I added the non-variant 'abon' via checklist (sorry for that) but that also would not allow me to add nonlisted values (was trying to add 'abondi', 'abondio' and 'abbondio'). Maybe it's something having to do with my account permissions? Or position of the ISS relative to the WeRelate servers? Thanks for thoughts. --ceyockey 02:11, 11 June 2023 (UTC)
I think I understand what's going on. The name variants project has a list of around 200,000 surnames. If you add a name that is not in the list (a "rare" name), but it has the same Soundex code as the name you are adding it to, then it is ignored, because rare names with the same soundex code are already included in search results. The problem is it should *tell* you that the name was ignored on purpose because it is a rare name with same soundex code. I'm sorry for that problem. I will correct it in a few weeks (I'm not at home right now) and let you know.--Dallan 08:08, 12 June 2023 (UTC)
Ahh. That makes sense. Thanks for rooting that out. Quite useful to know. Regards --ceyockey 01:33, 14 June 2023 (UTC)

Non-Latin characters, i.e. diacriticals [5 November 2022]

I noted that the add text box (3rd column 'Confirmed Variants') will not support addition of, for instance, ä or å. It there an intention to extend the accepted character set to include characters bearing diacriticals? --ceyockey 00:59, 6 November 2022 (UTC)


Cognates vs. variants [5 November 2022]

For 'janssen' I've added 'johnson' as a variant based on 'janssen' being the Dutch cognate of English 'johnson'. Is this type of relationship within scope of the variants project, or should I remove that relationship? --ceyockey 01:01, 6 November 2022 (UTC)


Bi-directional relationships? [6 April 2023]

Given a surname and a variant is added, is the relationship bidirectional? In other words, starting at 'janssen' and adding variant 'jansen', will the search for 'jansen' return 'janssen' as a variant, or do I need to go and add 'janssen' as a variant of 'jansen'? Thanks for clarifying. --ceyockey 01:06, 6 November 2022 (UTC)


Duh - it does appear to be bidirectional. Looking at the 'jansen' entry shows 'janssen' as a variant. I'm no names expert, but full bidirectionality might not be good for some relationship types, such as known precursor-successor relationships where the precursor is no longer in use. Just thinking out loud. --ceyockey 01:10, 6 November 2022 (UTC)


Bauer [14 April 2023]

There is a given name "Bauer", which doesn't make any sense. In my opinion it should only be a first name. --Flominator 07:00, 6 April 2023 (UTC)

Thank you for the feedback. I agree. I plan to make some improvements to the variant names project (improving the computer-generated names) later this year. I will re-visit the given vs surnames at that time.--Dallan 15:33, 9 April 2023 (UTC)


ß? [14 April 2023]

I would love to add Hanß and Hannß to the variants for first name Hans, but I guess it won't let me. --Flominator 17:47, 14 April 2023 (UTC)

I'm pretty sure ß is converted automatically to "ss" so adding Hanss and Hannss should work.--Dallan 23:54, 14 April 2023 (UTC)

On-page variants - remove them if "low value"? [13 May 2023]

Wondering whether there's support for removing low value variants from surname pages. For instance, for Surname:Harvey there are five variants listed all with source "WeRelate - similar spelling". Would it be OK or proper to blank the variants field in this case? Thanks for the guidance. --ceyockey 00:42, 30 April 2023 (UTC)

I think that's fine, but the list of variants that's actually in use is the Variant names list: https://www.werelate.org/wiki/Special:Names?type=s&name=Harvey
I'm not sure of the best way to communicate that the related names on surname pages aren't used any more - that they've been replaced by the variant names project. I'm open to suggestions!--Dallan 05:27, 30 April 2023 (UTC)

A suggestion: At left is how the "related names" box appears on Surname:Hatch after removal of the 'low value' names - all were of that type. On the right is a suggested replacement, an "Auto-links" box with links to

Image:Screenshot 2023-04-30 122313.png

This would leave the related names box intact but disable it's output, effectively hiding it. Over time, could survey a) % of the related names boxes have content and b) new content added to the related names boxes. After some threshold is passed of low use and low new use, could remove the functionality altogether.

--ceyockey 16:29, 30 April 2023 (UTC)

I like it! I will add it this weekend.--Dallan 04:11, 2 May 2023 (UTC)

I just made the changes. What do you think?--Dallan 23:15, 13 May 2023 (UTC)

  • Looks good. Thanks, Dallan. --ceyockey 01:46, 14 May 2023 (UTC)