[OSM-talk-ie] Typos in Townland names

[Brian Tuffy]

Today I hope to upload my list of typo changes for Mayo to OSM.
When Rory runs his Logainm script tomorrow, I hope that it will add more
logainm tags because of this, let's see. Most of the typos are just
alternative spellings, The list here are just the ones that I plan to
overwrite. There could be a valid local spelling of a name that I have
missed. I challenge you guys to check my "Typos" column for a valid
townland name spelling.
https://docs.google.com/spreadsheets/d/1wobpofwXzLdkPyB8mjMT1sDMQuZiO
gebR4S94yvbzU4/edit?usp=sharing

I manually checked the names to make sure I am not overwriting valid names.
I learned a lot about the structure of the names, false positive (i.e. name
mismatches that are not typos but valid alternative spellings or
alternative name structure)... and so I wanted to share those with
everyone. I hope this will help others who attempt to correct typos, and
help those working on automating the process in the future.

My assumptions:
(1) I assume that the GSGS3906 1:25k War Office map sheets, are the
original source of townland names and the spellings are valid names.

(2) There can be multiple valid names for one townland. Valid name(s) can
be from: the map source, a structure variation (e.g. X East, East X, Upper
X Island West More, etc.), a local/commonly-used name (e.g. Bellavarry is
commonly written as Ballyvary, both are valid and so should not be
overwritten),

(3) With the correct spelling of townland names, the Logainm script should
be able to automatically add logainm tags. A lack of logainm tags is an
indication that something is up with the name. After the typos are fixed,
any remaining townlands without tags can be added manually. Islands are not
included, yet.


Guidelines:
(A) Make double sure that the typo is a mistake and not a valid alternative
name as I outline below.
(B) Try to keep the names where possible by adding alt_name, loc_name or
old_name tags.
(C) Try to keep consistent naming schemes. For example, If there is an "X
More Island", then also try to keep "X Beg Island" and not "Xbeg Island",
Its not always possible though, I would not create a new name just to be
consistent.
(D) Its sometimes ok to keep invalid names, if they are common spelling
errors and there is proof that it is in use. For example, "Srahcorick" gets
a few search hits. "Srahacorick" is the correct spelling but I recommend
not adding the alt_name in this example. An example where I would add an
alt_name is "Bohea", official correct spelling is "Boheh". Local names
often come from spelling mistakes that become popular.
(E) The name= tag is the one that is usually displayed/dominant. I suggest
that the order of "importance" for this is local name > other names. If
someone knows a local or commonly used name of a place, they should use
loc_name=, but often they just replace the name= tag. (This is all
debatable, which name tags to use? alt_name is for townlands that have 2
official names. I also use alt_name for valid spelling variations. name:en
may also be used etc.)


How to identify Typos:
(1) Compare the list of townlands to the openOSI and other sources. Logainm
(derived from OSI), local authorities, etc.
(2) Visually identify them in your area,
http://maps.openstreetmap.ie/os_opendata.html
(3) Missing Logainm tags indicate a name mismatch.
http://umap.openstreetmap.fr/en/map/townlands-missing-
logainm-data_83833#11/53.9650/-9.2107


How to check that the "typo" is a mistake:
(1) Check if it is an alternative spelling or alternative name. (see below)
(2) Check the map source.
(3) It may be obvious why the typo occurred. Is the mis-typed letter hard
to read?, blocked?, or a common mistake? (See common mistakes). This is
only an indication but it helps.
(4) If the name on the map source does not match any in the relation (even
as alt_name), this indicates that it was misspelled when first copied or
possibly overwritten.
(4) To check if it was overwritten, check the relation history on OSM, if
the name did not change at all, it is likely that it was misspelled when
first copied.
(5) Do a web search for the name (be sure it is the correct townland), if
there are lots (> 50 maybe) of webpage posts with that spelling, no matter
how wrong you think it is, it is likely a valid name spelling. Be aware
that some websites just use OSM data so there will always be a few hits.


Alternative names or 'typos that are not typos':
(A) Many townlands have two official names on the map source, "Ballindoo or
Doocastle". Both are valid, We should try to find out which one is
commonly-used/dominant and use that one in the name= tag and the other in
the alt_name= tag.

(B) A name structure variation is valid, "East X" = "X East". We don't have
to add every variation but if you come across them, you can add them. The
idea is that if someone searches for a townland, we want them to find it no
matter what variation they use. The example above is trivial for search but
something like Inishmore Vs. Inish More.
Also consider, Upper/Lower, Island, More/Beg Oughter/Eighter etc.

(C) Missing spaces can be valid names. e.g. Inishdeashmore =
Inishdeash-more = Inishdeash More

(D) Island's have on island name= tag (place = islet) and the townland of
an island has a name= tag too (place = locality). These should match but
sometimes differ. e.g. Cullenmore island (Islet/island) and Collan More
(Townland).

(E) X Deme(ns)e and X Deme(sn)e. I suggest sticking to Deme(sn)e as this is
what is on the map source.

(F) Identical townland names are common, many appear beside each other
(without north/south upper/lower etc.). check the source to see if they
should have a secondary name in brackets after them. For example,
"Dookinelly (Calvy)" and  "Dookinelly (Thulis)". Sometimes names in
brackets after a townland indicate which civil parish it belongs to, but
not always.

(G) Some townlands are named "Part of X" or "X (part of)". This means it is
a separate part of a nearby townland. Usually, you will find an "X"
townland nearby, that was split.
I suggest we write the name of these as "X (part of)".

(H) Apostrophes in names. I'm getting a bit too in-depth here but I might
as well mention it. The old maps do not contain apostrophes in names.
"Georges Island", "Abbots Island". Some name tags are now "George's Island"
etc. I suggest to keep "Georges Island" as an alt_name in this case.




Sorry for the long email, I will try to put this into a wiki or something.
Please check over my list and see if you can identify a spelling that
should not be overwritten.
More importantly, any suggestions on the above text are also a good idea.

Brian T
OscarBrownbread




On Tue, Feb 14, 2017 at 3:23 PM, Brian Tuffy <brian.tuffy at gmail.com> wrote:

> Hi all,
> I found a lot of typos in townland names in county Mayo and i presume it
> is similar in other counties. As of 12.01.2017, About 4.3% of townlands in
> Mayo have a mistake in the spelling. That is roughly 130 townland names
> just for Mayo! I think that is something important we can correct now that
> the hard work of mapping the townlands is done.
>
> Before, I was correcting them as I came across them but I think a more
> systematic approach is needed to double check names and also why i want the
> community to help, share ideas on how to do it and check each others work.
> It is understandable that mistakes were made since it is sometimes
> difficult to read the GSGS30906 war maps source. So what I am talking about
> here, is spelling mistakes in copying the townland names from the source.
>
> So here are my questions;
> How do we go about cross-checking the names of townlands (also parishes,
> DED's etc)?
> What sources can we use? 1911 census list of townlands? the war maps can
> still be the original source of townland names. Can we legally use the data
> on pobal.ie to cross-check spelling? Although unlikely, can we even
> extract the other data there (townland id)? Is that useful ?
>
> Are you interested in checking the names in your county?
>
> Here I propose my way of checking townland names:
> The principle was to compare the OSM list of townlands to other lists and
> identify if the names match or are unique entries. The unique entries are
> possible mistakes.
> I downloaded the townland list for Mayo from townlands.ie and I am
> currently trying to compare the list to other sources. I have the draft
> list of townlands in Mayo from Loganim.ie and a townlands list from
> pobal.ie (possibly copyrighted list ). Anyway, once the possible mistakes
> are highlighted, (Excel - conditional formatting) we can go back to the war
> maps source and check the correct spelling. It takes a lot of work, i.e.
> formatting, false positives, we have to respect local names of townlands
> etc.
>
> Besides typos, this method can also be used to add the irish names of
> townlands, identify missing townlands not yet added and much more.
>
> Please let me know what you think. My corrections for Co. Mayo are nearly
> complete and I plan to upload them soon.
> Ok, Thanks for reading,
>
> Brian
> Username: OscarBrownbread
>
>
>
>
>