Proposal:Changeset tags

Changeset tags
Proposal status:	Draft (under way)
Proposed by:	Sletuffe
Draft started:	2012-10-21

Following Richard's suggestion at [1] this page aims at proposing a set of recommended or mandatory tags to add to changesets so we can agree on some standard and reproducible way of identifying by human and computer means types, sources or other informations characterizing edits.

Wherever and whenever possible, editor softwares (or plugin software) should try to add those tags automatically.

See Changeset#Tags_on_changesets for actual changeset tags being used nowadays.

tags

tag name	value	description	example	mandatory ?
comment	* (free form text value)	This is the de facto standard way, implemented almost everywhere for the contributor to add a human readable description to his edit	"Hiking with my GPS, bad reception in woods, but otherwise should be good quality" "Corrected errors found with keepright"	recommended
created_by	* (free form text value, should be automatically added by editors, please try to use a software usable string)	This is the de facto standard way of indicating the software used to add data and it's version	"JOSM/1.5 (5181 de)" "Merkaartor 0.18 (it)"	highly recommended (the editor should do that automatically for the contributor)
source	* (But try to find what others have used with the same source to respect a common syntax)	This tag is to be used the same way as source=* on objects and may be a future replacement to object level source tagging, when using it, it is highly recommended that your edit should only contain data from that source.	"Bing Aerial imagery" "gps" "A mix of survey/gps/bing" "ABC institute 2011 mission"	Highly recommended when you are using whatever external source when contributing data (aerial tracing, GPS, survey). In case of imported data where vertices are built from another source, this becomes mandatory.
source:license	<machine-readable licence name>	Use this to give the license of the vector imported data	"CC-0" "ODBL" "Public Domain"	highly recommended when type is set to import
source:date or survey_date (see stats)	<ISO date>	date of survey, last-modified of import data, date of out-of-copyright historical map, date of Bing/Landsat photography etc.	2010-12-31	optional (useful for older sources)
imagery_used	semi-colon separated list	A list of background imagery sources (Bing, TIGER, OSM, etc) used while editing. Unlike source, imagery_used does not imply these backgrounds were the source of the knowledge, merely that they were shown.	Bing;MapBox	may be added by editors automatically
imagery_offset	two offsets separated by ";" for single Imagery, see also source:geometry=* for per-object tagging		Example: 13.43; 8.42 Example2: -3.43; -8.42	may be added by editors automatically
bot	yes	Use this when the performed edit was entirely done by a bot without human control during the whole operation	(If an editor like JOSM or Merkator was used for the operation this shouldn't be used, use type=automated instead if that apply to the edit, see below)	Mandatory
type	import	Use this when the data you are integrating into OSM contains vector data from any other source. (If you plan on correcting manually this data, you can still do it before uploading your changeset but that doesn't change the type=import requirement. If you plan on adding other related data by other means (aerial adjustment, deduction) please create another changeset.		Mandatory (Such a JOSM patch can help making it mostly automatic and less forgetable : [2] when the source format is an .osm file)
type	automated	Use this when the performed edit was done in order to modify data where not every single change was looked at (but not by a fully automated process)	This especially apply when someone download a bunch of object with common properties, and then change all of them in JOSM	Highly recommended, might change to Mandatory (A prior discussion with the local or global community is Mandatory if such an operation would touch objects not only previously touched by you)
type	revert	Use this when the performed edit was done in order to revert some other edit, in part, or as a whole		Highly recommended, (the tool/plugin/editor used for the revert should automatically set this)
reverted_changeset	<id>	Use this when the performed edit was done in order to revert a whole other changeset (<id> is the id of the changeset)		Recommended when the revert was applied to a whole other changeset
type	tracing	Use this when the performed edit was done by interpreting a background image and tracing over it (Aerial imagery tracing, or other backgound image of any type, may it be raster images or raster images built from vector data) Adding the source tag make is really interesting to evaluate what is the expected quality or age of the data wich was used.		recommended
type	fix	Use this when the performed edit was done in order to only fix other contributors errors reported by QA tools		Nice to have if those fixes are performed with a dedicated editor like rawedit, otherwise it's just common to not set any tags for that.
type	survey	Use this when tracing mostly from a local survey (with or without GPS traces)		Really optional, as it is often considered as beeing the fail over if no other type=* is set, but If you want to explicitly say that is was a survey, you can use this tag
transport	*	?? (perhaps the mode of transport used while conducting the survey?)	"car" "foot"	optional
url:<language code>	*	Use this to link to further explanation of the edit (link to the wiki.openstreetmap.org explaining page prefered but blog, personal page, mailing list post still ok) le language code indicates what language for the page to expect.		Highly recommended when bot=yes or type=automated/import

sample current statistics of used tags on changesets

The following are the changeset tags used at least 100 times as of October 2012.

created_by           | 12488949
comment              |  9882933
version              |  2013859
build                |  2011697
bot                  |   317744
source               |    85151
create_by            |     6607
redacted_changesets  |     4046
created by           |     3812
survey_date          |     1097
transport            |     1024
survey_type          |     1015
CLC:content          |      880
CLC:year             |      880
created              |      734
is_in                |      725
source:date          |      655
note                 |      529
type                 |      517
is_in:country        |      473
(new key)            |      411
import               |      324
author               |      249
Source               |      210
user                 |      202
source:name          |      201
source_url           |      193
source_ref           |      186
bot_url              |      185
osc filename         |      170
is_in:state          |      167
reviewed             |      140
url                  |      110
merged               |      104
uploaded_by          |      102
name                 |      100

here [3] is some interpretation of their meaning. (credits do P. Norman)