Suggestions Report Suggestion - Locations

Question

Suggestion for the suggestions report. The Profile Completeness suggestions are only reported when specifically requested by the profile manager. Suggest that the following be reported only for orphaned profiles:

• 457 Profile completeness - Birth Location not set
• 459 Profile completeness - Birth Location Country not recognised
• 460 Profile completeness - Birth Location Country not official
• 463 Profile completeness - Death Location not set
• 465 Profile completeness - Death Location Country not recognised
• 466 Profile completeness - Death Location Country not official

Answer 1

Agree, but dont really understand why we dont just include these in the suggestions for everyone. These help with matches, with merging, with searches, etc. Anybody who wanted to do any mapping of locations would need this too

Comments

See this thread https://www.wikitree.com/g2g/959348/question-week-what-your-wishlist-2020-wikitree-improvements?show=959348#q959348

---

Please don't make these mandatory on all profiles, at least not in the suggestions report.  Even the ones that individuals have turned on have clogged up my extended relatives suggestion report that it's harder to find the real errors (even with the color coding) and discouraging.  I would rather have them as a separate report even for those optional choices.  These are more of an improvement report than an error report.  Often these locations aren't known because the church or government records were burned, so they might never be known for sure.

---

Cindy, these errors are saying they are missing entirely, or are incorrect at the Country level. Surely that much can be at least inferred from other sources/family members? There may be a few immigration/emigration issues that would be mistaken, but the vast majority would be easy fixes.

---

Cindy -

This is for completely missing BMD, preferably when none are supplied, and then only on orphaned profiles. Yes, it's an improvement, but then again aren't most of the suggestions? This of this as a parallel to the 131, 132, 133, 134 suggestions.

Aside: On the tags I follow, including a couple common surnames like Reed, Whiting and Sands there is usually a new profile created daily that lacks any location.

It's an improvement that helps us identify/preclude duplicates and to resolve some of those more difficult situations where profiles are conflated but geographically infeasible.

---

I was talking about Jonathans thought that it should be expanded to all profiles.  I estimate on the 25,000 profiles on my extended suggestions list, if even a quarter of them are missing even one of these 6 items, I'd have 6,000 more suggestions.  That's overwhelming.

Answer 2

After having a look at the table it is very incomplete and contains errors for the locations I mostly work on. I'm going to prepare a list which I will email to your project leader.

I'm afraid encouraging members to modify the locations on orphaned profiles may result in adding incorrect locations to profiles which were previously correct. (Please do not misunderstand me. The majority of fixes will probably improve the profiles, but there will be a measure of incorrect fixes).

Comments

You are probably talking about https://wikitree.sdms.si/function/WTShowTable/Table.htm?table=Countries

By all means. You can post it here or email me or whatever. Last week I added departments for france based on geoNames database to improve geolocation of the profiles. The way they are set now will improve geolocations of many profiles and shouldn't produce any additional suggestions.

---

Indeed, that is the table I was talking about. Thanks. Departments should only be used from 1790 on. Before that we had provinces. And some places now in France where not in France before, for instance Savoie only became French in 1870. Here is a page with a list of Provinces and a general idea of how they match present-day departments: https://www.wikitree.com/wiki/Space:France_from_Provinces_to_Departments

It's not organized in a table with date of start/end, but I can think of building such a table, if it would be useful.

---

It is up to you into how many details you want to go.

We can include timeframes or not. If we do, then eventually we can make suggestions for invalid entries.

Table would definitely be useful for me. And the table should have geoNames administrative code to make geoLocation of profiles better. I can set it up like I did for England.

---

The real solution would be a GIS based system. So name, start-end, geo coordinates. Then the name is enclosed by a larger one, also depending on date.

See eg Radisch (2019) Automated Geo-resolution of Place Names in Historical Serial Sources. Proceedings of the Doctoral Symposium on Research on Online Databases in History 2019. 42-46.http://ceur-ws.org/Vol-2532/paper7.pdf

---

That looks useful, but it is only an algorithm that I already did implement in my algorithms. But there is no mention of the data being available for the whole world.

---

No, the data for historical location referencing with names... world wide... that is as big a project as WT I bet (hence the 'would be'). But I would need to know more about your dataset/algorithm to know how far along your are.

I found one old projects trying to do it: http://commoneras.ecs.soton.ac.uk/contact.html

https://sourceforge.net/projects/chalice/

See the discussion here: https://forum.geonames.org/gforum/posts/list/20/1242.page

---

My approach to geolocation is based on the location table. I try to find the match of the location ending defined in my location table. Each item in location table is associated with geoDatabase admin areas, that cower that region. Timeframe is ignored for now, but will be implemented in the future.

If a location ending is matched, then I search the remaining part of the location in geoNames only in defined admin regions. And the best match is the resulting location. If the ending is not matched, I search for each part of location in the geoNames and intersect the results. This has a tendency to put a location on the other end of the world, since place names often repeat and there is no way to know which "London" is correct. I do have some conditions to get the best match, but that is still being developed.

---

Isabelle, where did you get this maps?

https://www.wikitree.com/wiki/Space:Cartes_des_provinces_fran%C3%A7aises

---

Aleš, they are from Wikimedia Commons. https://fr.wikipedia.org/wiki/Utilisateur:Oie_blanche?uselang=fr

---

I added Berry as an example.

https://wikitree.sdms.si/function/WTShowTable/Table.htm?table=Countries#FRA

---

Isabelle,

Yes, any suggestion runs the risk of an incorrect correction. I am envisioning something like the 133 suggestion. Most of the ones I have encountered (looking at old GEDCOM imports) have entire families missing locations (and also lack sources) -- or some cases where the mother has a location for birth and death but none of the children do. Even something as broad as United States, England, Europe would help us discern different people ....

Isabelle/Ales,

It has been ages since I have looked at the GNIS. I do not know what the requirements are to obtain/use this data, and seem to have lost my account info. I do remember that the US government spent over a decade to gather this information. It is continually updated, including the validation of the sources for data updates.

From some of the GIS research that was active when I was working in this area in the 1980s - dealing with the temporal nature of the data is a challenging problem.

---

That is a very interesting table Aleš.

I see you have one town in the Netherlands at the top level for the country (Gouda). That links to spelling variations, which are then searched for in names/locations/bios?

I assume you have linked those to the Geonames ID? Then getting the geo location shouldn't be too hard (well, centre point probably).