ERROR REPORT - Placename standardization
@N Tychonievich please and thanks
In the 1901 Ireland Census, a civil address (in County Meath), containing townland, parish, and county, has been corrupted to a Diocese and Province - an old-form church location. Not even the same kind of address/location.
Answers
-
Part of the problem is that the "Event Place (Original)" is not in the database because even when it's not misspelled, it appears to be the middle of nowhere. Either that, or the Townlands of Ireland website gives the wrong coordinates for it.
https://www.townlands.ie/meath/lower-kells/moynalty/trohanny/carrigagh/
A 5-km-radius search in FS Places doesn't yield anything even vaguely like "Carrigagh" or "Trohanny": https://www.familysearch.org/research/places/?includeIsParent=true&lat=53.8608&lng=-6.9384&mapShowCircle=true&searchTypeaheadInputText=53.8608,-6.9384&distance=5k.
Now, granted, the pin for the Diocese of Meath is in Dublin, 50 miles away as the crow flies, so it's clearly wrong, but I'm really at a loss for what to suggest instead. I guess I just don't know enough about Irish geography. :-/
0 -
Trohanny is a known place in County Meath in the Ireland townland database. https://www.townlands.ie/meath/lower-kells/moynalty/trohanny/trohanny/ Shifting it to County Dublin is nonsense.
Corrigagh is Corgreagh, in the same barony, Lower Kells. https://www.townlands.ie/meath/lower-kells/moybolgue/trohanny/corgreagh-or-killagriff/
0 -
"Shifting it to County Dublin is nonsense"
Well, yes but I would suggest that there are (at least) two issues.
Firstly Trohanny is clearly a known townland but it's not in the FS Standard Places.
Secondly, because neither Corrigagh nor Trohanny are in the Standard Places, then my personal belief is that the system just looks for "Meath, Ireland". There are two matches - the County and the Diocese. For whatever reason, in this case it's chosen the 2nd, and for some reason, the co-ordinates of the Diocese are in Dublin. Whether there is a reason for that, I don't know.
(NB - I hope I'm correctly remembering how Event Place (Original) maps to Event Place)
Incidentally, I wondered if Corrigagh was supposed to be Carrigagh - see (see https://www.townlands.ie/meath/lower-kells/moynalty/trohanny/carrigagh/ ) According to that site,
Carrigagh is in the Electoral Division of Trohanny, in Civil Parish of Moynalty,
That's a different Civil Parish but I couldn't get the census forms up that included a CP to check.
Presumably none of these are in the Standard Places. Isn't that what needs to be fixed first?
0 -
Ah - so it's MY fault for not fixing the place? A place I didn't know I needed to research before today.
Isn't the issue, in fact, that the algorithm was so poorly designed that it gets it wrong more often than right?
1 -
"Isn't the issue, in fact, that the algorithm was so poorly designed that it gets it wrong more often than right?"
An algorithm can only work with the data that it's given and I've seen several threads complaining about missing townlands in Ireland. Given that there are sites holding this data (like townlands.ie ) I'm disappointed that a download into FS hasn't been arranged. If that had happened then we wouldn't be marooned with every new event we enter (because yes, it's not practical for us to get the places right before we research the people).
Then there's the process of recording an Event Place (Original) on the Historical Records. That ought to be audited for the place being in Standard Places - if that had happened, then "Corrigagh" and / or Trohanny would have been picked up before people like yourself try to use the data. (I've not dared to enter my Irish data into FS yet - which means I have the luxury of messing up and correcting Irish place-names on my PC without affecting anyone else, as I was doing today).
0 -
BUT the algorithm was created by FS, by programmers who SHOULD know that their database is imperfect. Yet it was designed to shove every place into a box, whether or not that is the correct box.
I would like to point out that @N Tychonievich has asked for these reports to be made in order to push them up the chain. I tagged @N Tychonievich, but it seems everyone else has an opinion, whether or not it's useful.
0 -
Relax, we're all on the same side here... :-) Or at any rate, I think we all agree that the autostandardization algorithm is badly flawed and that its effects should have been rolled back years ago, instead of the piecemeal "fixes" that poor N Tychonievich dutifully sends up the chain.
This example just points out the futility of expecting a bot to produce anything resembling logical results when it's working with a faulty, incomplete database combined with all the vagaries of indexed data. Of course, Murphy's Law also applies: of the two semi-matches available to it (the county or the diocese), the bot chose the worse one. (The pin for the county is a bit south of Navan, about 20 miles from the townland.)
But hey, at least it's the correct country! Even having the correct continent is an improvement over where we started.
2 -
I'm hearing "Don't blame FS for making a mess. You should have fixed it."
I spend a GREAT deal of my time working in Irish records. Finding Mayo in Galway and Tyrone in Down on an hourly basis is not my idea of a good way to spend my research time.
0 -
@Áine Ní Donnghaile - if you're hearing that it's your fault, then I totally apologize. It was never my attention to suggest that you personally should prepare the place names before you start work. That's totally impractical.
The annoying part for me is that there are websites with lists of Irish townlands and yet I see no evidence that FS has used them. (Note "see no evidence" - I'm prepared to be shot down if something is happening behind the scenes that we're not being told about)
In this particular instance, there are two sides to the same coin - the algorithm and its data. In general, both sides have to be fixed for our events to encode correctly. Or approximately correctly. In this specific instance, FS doing a data fix will be a crucial part. There are other cases where it may be that the algorithm is the only thing that can fix things. I'm honestly only trying to help FS sort itself out by suggesting what FS can do.
1 -
@Áine Ní Donnghaile I'll send it up--and note that the standardized places need updated in re Irish townlands.
2
