Talk:Key:source

From OpenStreetMap Wiki
Jump to navigation Jump to search

Discussions held elsewhere

  • Tagging email list
    • January 2013, "Source tag - deprecated for use on objects?" (This discussion faded out quickly with most replies stating that it was not deprecated, no one could agree if it was worth keeping and the overwhelming majority agreeing that it was certainly premature to deprecate it at that time).

Automatic addition of Source value

Good Idea. Perhabs we schould also have a automaticly filled property, about how mad changes to a object. Like Wikiepdia "Version" tab. Sven Anders 13:43, 28 Jul 2006 (BST)

There is no suggested core value of GPS, is it assumed that if a node is above an uploaded GPX track that that is enough evidence of its source? Dmgroom 00:03, 5 September 2006 (BST)

I think it is assumed, but on the other hand I'd still add the appropriate source tag even in that case when creating the map. After all, some node/segment/way could be above a GPX track just because it's the same road, and yet it could come from a different source... using the appropriate tag in all cases won't hurt. LjL 18:25, 5 September 2006 (BST)

source:name

I actually don't like source:name since it does not comply to the logic used on other tags (e.g. all the lane proposals, the mtb stuff, and so on). I'd love to see a generic foo:bar:source approach, meaning that source:name would be replaced by name:source and the whole concept could be extended to virtually any tag, e.g. opening_hours:source, description:source and so on. I just find source as the last suffix more logical, since the value of the key is actually the source you describe, therefore it is just logic to place the source tag in last place. thoughts? --Marc 18:49, 29 January 2010 (UTC)

source = survey (wiki) vs. GPS (JOSM)

When I create streets etc from a GPS track, I'd like to note that - the wiki suggests using source=survey but JOSM does not offer survey but GPS. As long as JOSM supposes GPS it will be much easier to use it instead of survey. Please clarify and note in wiki (http://wiki.openstreetmap.org/wiki/DE:Map_Features#Erl.C3.A4uterungen and http://wiki.openstreetmap.org/wiki/Key:source) what to do / change JOSM proposals. This comment was not signed, so no easy way of knowing by whom or when this was said!

Personally, after finding these two values exist, I use both but with different meaning. I user source=gps when I actually measured the feature location with my GPS. I use source=survey when I walked within eyesight of the feature and observed its properties and relative location (for instance, I personally saw that the item was there and that it was a bit north of where I was standing, as opposed to being to the south of me, but I had no opportunity to actually walk up and measure its GPS coordinates). Jbohmdk 02:40, 10 June 2013 (UTC)
I have now added this to the page (as a separate edit in case anyone wants to revert this or one of the other edits I just did). Jbohmdk (talk) 23:46, 9 April 2016 (UTC)

Multiple Values

First Entry

(before structure was generated and so probably easy to be overseen after structure had been generated - so added to this chapter retrospectovely)

If I map a road and use many different sources, e.g. Yahoo Image, my own GPS-Trace (which I uploaded after using), GPS traces from other users and also data from http://wiki.openstreetmap.org/wiki/Import/Catalogue/Strassen_NRW . I use different sources in order to map the street most realistic, but with the source tag I can only declare one source. And some people even tag every road with source=survey. Is that good? What can I do / how can I declare more than one source? And if I use Yahoo Image for mapping, do I *need* to declare that I used them?

Thanks in advance. :) ([[[User:Westfa|Westfa]] 13:22, 1. Mai 2009; added retrospectively by --Rennhenn)

official way to assign multiple values for a tag is to separate them with a semicolon, thus you could tag it like source=survey;yahoo or whatnot. i'd say - don't bother too much with it, just set the one that contributed the most. though personally, i only add source when i have used some 3rd party source like landsat etc. --Richlv 19:02, 1 May 2009 (UTC)

Second Entry

If (say) a way has been traced from Yahoo! imagery (hence source=Yahoo), I go there and survey with my bike writing down information like lanes, surface, lit and so on but without a GPS track, should it be source=Yahoo; survey? --Nighto 00:21, 8 May 2010 (UTC)

See above, I asked the same question one year ago :D Seriously, I think that the source key should only be used on an element (a way for example) if the way is very different from what it was before or if you use keys and attributes, that seem "extraordinary". If I map a normal street I don't set the source key on the way, instead I write in the comment of the changeset what I did and where the information come from. And if I used 3rd party sources I add a source key the the changeset. That only works of course if I only use one source. If I said source=Yahoo;survey;landsat;blahblahblah it's not clear for which element I used which source. But if I use different sources for different things I just use different changesets. And of course I add the source key to an element (e.g. node, way...) if another mapper could wonder why I set those values. Under [1] you can find an example how I use the source key on a changeset. [1] On the other hand Openstreetmap is *free*, so *you* can decide how it should be tagged. Other people say the source key on every way makes sense and should be expanded. Just do what you think what makes sense. It's easy to look for the comments and source tags in a changeset with josm, because there is a very nice history function implemented now. :) --westfa 20:31, 17 May 2010 (UTC)
Just put the source=* on the changeset. That way it is easy to see what information came from where. Gnonthgol 22:16, 17 May 2010 (UTC)

Is this tag deprecated?

I've been told on one of the email lists that this key is deprecated and that it should no longer be used on objects but should be used exclusively on changesets. Could someone please confirm that? --Ceyockey 11:57, 2 January 2013 (UTC)

The documentation on the main page has been revised (not by me) to reflect that this tag is to be used on changesets and not objects. --Ceyockey 12:48, 2 January 2013 (UTC)
I made the revert you refer to. I had intended to return a couple of hours later and make some more changes and provide something more neutral, but I see that Frederik Ramm has already made some of the changes I intended. The tagging@openstreetmap.org list currently has discussion going on over this issue, and I assume there will not be agreement. I will state with confidence that adding source tag to the object is not depracted because there is no means of doing this and it appears the majority of people use the source tag on objects. I might be worth giving some though to how your local/regional community uses the source tag. -- Jamicu 15:25, 3 January 2013 (UTC)
Which revert is that? You or someone else seem to have deleted whatever you were responding to, making this part of the discussion impssible to follow. Jbohmdk 03:39, 10 June 2013 (UTC)
I guess this one ;-) --rayquaza (talk) 18:37, 4 July 2013 (UTC)

A source tag on every object in a changeset is very impracticable. Mapper X used Yahoo some years ago, mapper Y refined from Bing 2012 and now we have even better precision from a local geoportal - now beyond moving the nodes, I also need to change the source tag on every object. Thus, having it once in the CS makes it much clearer when which imagery was used.

The exception are objects where other sources have been used compared to the surroundings, such as a new building surveyed by sophisticated GPS techniques. In this case the source tag also works as a warning that it might not yet be visible in the aerial.

Absolute nonsense is the source=HiResAerialImage that I see a lot recently, without saying where from and which year. Now that local councils make their geoportals accessible it is quite important if that was traced from high res Bing 2012, high res local 2011 or high res local 2014. --Polarbear w (talk) 15:58, 17 December 2015 (UTC)

My changesets may have up to 4 sources. I add source tags to the new objects in order to identify the particular source used for that object. I don't see any harm in it other than data bloat. Warin61 (talk) 23:47, 4 August 2018 (UTC)

Setting chapters to "historic" - changeset vs. object

There was a recent effort by Jbohmdk to set several chapters on this page to "historic". While I agree that the use of the source tag is moving towards the changeset, it still has its place on individual objects, and in particular giving the source of a particular attribute, such as source:name or source:geometry is absolutely valuable and useful information. So they are not "historical" at all. They are important if the source differs from that of the changeset, or from the surrounding objects. E.g. while all the old buildings can be traced from aerial imagery, a new building could be surveyed by GPS, and it is important to emphasize this. Further some attempts to define values are to narrowing, e.g. there are different forms of survey. --Polarbear w (talk) 08:24, 11 April 2016 (UTC)

Please group the common values for this by scope

Currently, the list of common values is a meaningless mishmash of robotic import source tags (each of which necessarily occurs millions of times) and tags that human mappers may actually want to use. It would be much more useful if the two were separated out from each other in this wiki documentation (but NOT as separate tags). e.g. first a list of common source tag values by human mappers such as survey, gps, local knowledge, bing, etc. Then a list of robotic tag values for bulk imported data.Jbohmdk 03:47, 10 June 2013 (UTC)

I agree. -T99 (talk) 07:36, 21 March 2015 (UTC)
Done (Though I am unsure if a few of the specific source for human mappers are actually robot sources added to the wrong place in the old tables) Jbohmdk (talk) 23:42, 9 April 2016 (UTC)

Survey vs. land survey

A "survey" can mean a lot of things. It can be quite informal ("I drove around the city surveying street name signs to verify TIGER data"), or very professional "land surveying" ("Our licensed engineering team set up our theodolites and laser levels and triangulated the exact position of the node with respect to existing boundary markers with a margin error of less than 0.03 m").

I am not a licensed land surveyor. When I say source=survey, it means that I was there, saw the object and took a note of it. I guess most OSM contributors are not licensed land surveyors. But some undoubtedly are. It would be nice to have a documented way to tag the type and quality of the survey performed. --T99 (talk) 08:18, 15 March 2015 (UTC)

I'd like to add that if you was on site and took a picture, "source=photograph" is really just a means of recording the information that you saw. The same applies to "source=audio" (or was it "source=voice"?) and "source=video" which I think were mentioned earlier but which seem to be forgotten now. On the other hand, if somebody else took the photograph (or video etc.), then "source=photograph" would be better than "source=survey". What if it's an old photograph you took before you started mapping? Or if the photograph reveals details you didn't notice while taking it?

Conversely, if your survey consists of reading signs (e.g. street names), then the survey is just a method of recording the information but the actual source of the information is the signage.

If you want to say "source=survey" accurately, I think you should have made at least some assessment of the object in person and on site, if only just "this way is a residential street in my judgment" or "that building looks like a house".

--T99 (talk) 08:17, 21 March 2015 (UTC)

Bing 2012

This sentence makes little sense to me: "Note that it is still 2012 and not 2013 because of the age of the images." What does it mean? Please elaborate.

It's actually 2015 now and most of the Bing imagery in my region is from fall 2013 ("9/7/2013-10/24/2013") or ca. 2011 ("11/3/2010-4/29/2012").

--T99 (talk) 07:48, 21 March 2015 (UTC)

The year should be the year in which the images where taken. The sentence was added in 2013 and needs to be changed. --rayquaza (talk) 16:44, 19 April 2015 (UTC)
The use of the year that the image was take assumes the user knows that. I would rather see the time/date when the source was added and the user that added it. This could be automated!!!! Or even similar to what is used here to add that information. Would make it easier to immediately see if the data is outdated (for those that know) and contact the author. And it would apply to ALL data sources, not just imagary. Warin61 (talk) 23:48, 5 December 2015 (UTC)
Now that the source tag is only supposed to be on the changeset, not the object, the user and upload time is already there for new uses. Jbohmdk (talk) 22:11, 9 April 2016 (UTC)
You can see the image capture date for Bing images in JOSM by selecting "Show Tile Info ..." in the context menu (right-click on the canvas). T99 (talk) 19:54, 16 June 2018 (UTC)

I have finally rewritten the section. T99 (talk) 19:50, 16 June 2018 (UTC)

USGS

For lack of any other documented tags, I'm using source=usgs for information acquired from United States Geological Survey. This includes topographic maps (through JOSM USGS Topographic Maps imagery layer, downloaded from USGS or traditional hardcopy) as well as Geographic Names Information System (online lookup or file download). I will add this to the table on the main page unless there are objections. T99 (talk) 22:32, 18 November 2018 (UTC)

How legal is the addition of data from social networks?

I surveyed a location and added its Instagram id. To check it, I gone to Instagram's site in the location's profile and there was the useful content of a possible opening_hours=* tag. However, is it legal to gather data from Instagram? I was already thinking on uploading just the opening_hours=* info on a separate changeset and putting source=instagram on it. It would be verifiable because the Instagram id is right there. (Unsigned post of User:Dhiegov at 4 March 2019)

It's a tricky subject. I believe that content on instagram may be considered a database, and therefore protected in some jurisdictions. This would make it not ok as an OSM data source. The problem is that while individual facts are not copyrightable, collections of facts (phone books, maps, etc.) may be – and even if any given mapper only copies a few opening hours, our community as a whole could pass the threshold and be at risk. So being big enough to qualify as a database of opening hours is what may make copying from Instagram different from copying a single set of opening hours from a regular website. IANAL, and this could be completely wrong.
If there is a legal issue, though, adding source tags is unfortunately not going to solve it. (It may help your fellow mappers verify the correctness edit, but that's separate from copyright concerns.) --Tordanik 19:03, 22 March 2019 (UTC)
Got it. So better leave it out of OSM just for precaution, right? So, better to personally contact the establishment operator to get that info. --Dhiegov (talk) 12:57, 23 March 2019 (UTC)
And it happened again, but not on social networks: I was going to map a healthcare=laboratory, but when I gone to their website there was a very convenient source of all their units, their addresses, phones and opening hours, but when I scroll to the bottom of the site I see an "All rights reserved" copyright notice. Just for reference, that's the page: [2] (it's in portuguese). Do their copyright notice covers all the data from their website? Can I get a permission to use this data by email? Because I think that if they are exposing it this way, they want people to know it and it would be stupid to not help who's helping them in this case. I'm afraid this website, by the same reason it's so convenient, classifies itself as a database :/ --Dhiegov (talk) 13:51, 30 April 2019 (UTC)
Contacting the copyright holder and getting explicit permission is a good way to bypass such legal concerns, of course. If you want to reach more people than just the few who happen to have this wiki page on their watchlist, you could always discuss it on the legal-talk list or ask the Licensing Working Group, but they'll probably give you a cautious "when in doubt, don't use it" answer, too (understandably). --Tordanik 20:08, 9 May 2019 (UTC)

source=photograph or source=image?

On this page the tag for photographs as source is named source=image. On other pages source=photograph ist used. Which is correct? --Thosch66 (talk) 10:26, 23 May 2019 (UTC)

Primary sources on the web

For e.g. changeset 75960904 which updates website, the source is the website itself which makes a claim to a street address. What should source be? --Andrew (talk) 18:10, 22 October 2019 (UTC)

source on elements

Do I read this correctly that source should only be used on changesets? I use source sometimes (additionally) on tags where the the source is not obvious. E.g. amenity=shelter + name=Some Name + source:name=guidepost nearby when the name is not written on the shelter but on the guidepost. I found/find this handy, because the next mapper still can verify it. Or should I then add source:name=guidepost nearby additionally to the changeset? --Ze0zohk1 (talk) 08:24, 31 December 2020 (UTC)

I continue adding source as tag for the most surprising or complicated tags, where it is likely that other mapper would mistakenly remove them. I do not add them if it provides no highly useful info Mateusz Konieczny (talk) 08:39, 31 December 2020 (UTC)
Same as Mateusz Konieczny, with the addition of trying most of the time to split my edits with 1 source = 1 changeset, therefore no need to add source=* or source:*=* on objects. --Gileri (talk) 09:51, 31 December 2020 (UTC)
I concur. I generally do bigger change sets, encompassing all recent surveys for a specific area, and try to keep the sources separate (for example, I try avoid mixing GPS surveys of brand new streets with detailed geometry traced in older aerial imagery in the same change set). But sometimes you encounter an item where some essential information is available only from some other source (for example, the name of a new school that you can find on the school district's website or a newspaper article). You could add that info later in another change set, but if you don't want to risk forgetting that, you can add it immediately with a source tag specific to that item. -- T99 (talk) 09:46, 31 December 2020 (UTC)
which software do you use? with josm, continue as you do, then at the time of sending, search for "modified source=survey", remove the tag source=survey, send the objects selected with the changeset tag source=survey. do the same for source=B, C etc. it makes the history much more readable to separate what is seen in the field (and is therefore totally up to date) from what does not come directly from the field and is therefore potentially out of date --Marc marc (talk) 13:14, 1 February 2021 (UTC)
That's what I do for fixes coming from Osmose, it adds source=* on objects (technical limitation of JOSM remote), but I remove it and put it on the changeset. --Gileri (talk) 13:29, 1 February 2021 (UTC)

Maybe this tag should be completely nuked from the database

I couldn't see any benefits of using this tag. -- CBRS (talk) 01:55, 2 March 2021 (UTC)

Please be aware of Automated Edits code of conduct before doing this. Mateusz Konieczny (talk) 08:56, 2 March 2021 (UTC)
Stating a source such as where you got information about a park or place you've mapped that might be helpful to other mappers isn't a benefit? --Lectrician1 (talk) 02:40, 3 March 2021 (UTC)
In most if not all cases source=* can and should (imo) be set on changesets and not on objects. --Gileri (talk) 20:20, 3 March 2021 (UTC)
Yes, in most cases it should not be added on objects (though for example all highway=proposed if are put into OSM really should carry it). Mateusz Konieczny (talk) 06:47, 7 March 2021 (UTC)
It would hinders using multiple sources pertaining to the same data (node positions, attributes etc) over time. Adding a source:* for each attribute and source=* on each node would be a headache. I think setting source=* on the changesets would still be more clear and efficient --Gileri (talk) 16:44, 7 March 2021 (UTC)
I agree that this tag on objet is a quality issue (the source on a objet doesn't said that the current version is using this source, the added infos from this source may have changed to the point that nothing comes from this source anymore, unlike the changeset tag which correctly links the source to the version concerned)
but in the same time, we approve proposal with suggest to add source on the objet (and in this case, incompatible source, wonderfull !) Marc marc (talk) 14:11, 18 March 2021 (UTC)
It is just ridiculous. How could it get passed? They are trying to bring garbages back in a sneaky way. -- CBRS (talk) 20:02, 1 April 2021 (UTC)

proposal: source tagging strategy should be decided at country level

I would like to propose the tagging source strategy (per changeset, per object, per object attribute) should be decided at the country wiki level instead of globally.

The wiki terms are confusing ("historic", "deprecated") and in most cases, new and existing mappers are simply following the old local way.

Personally, I think tagging at changeset level has little value:

  • you cannot prevent a mapper to use multiple sources, hence there will be no way to know which source was used for each object => counter-intuitive
  • information is impossible to access in ID editor, you need to navigate to each object external OSM link and then again to its changeset link => counter-productive

You may not agree and have valid arguments and it's fine but I believe the wiki should leave the choice to each community to decide what is best for them.

I agree that some things are best adressed on a country-by-country level, as there are a lot of variation on the ground. However I fail to see how a source metadata field falls into that category.
From afar it looks like a "divide-and-conquer" tactic, if I may.
>you cannot prevent a mapper to use multiple sources, hence there will be no way to know which source was used for each object
The same issue is present wherever you choose to put the source tag
> information is impossible to access in ID editor
Just like mapping for the renderer, one shouldn't map for the editor ;) --Gileri (talk) 18:22, 18 August 2021 (UTC)
"I would like to propose the tagging source strategy (per changeset, per object, per object attribute) should be decided at the country wiki level instead of globally." - of some area has clear consensus they are free to do so! Note Relation:associatedStreet that is deprecated and being eliminated (eliminated?) in Germany and widely used in France. If such consensus is reached it can be documented here. Mateusz Konieczny (talk) 10:54, 19 August 2021 (UTC)
"you cannot prevent a mapper to use multiple sources, hence there will be no way to know which source was used for each object" - the same applies for source tag. Though source tag on object typically has opposite problem and is not actually listing all used sources Mateusz Konieczny (talk) 10:54, 19 August 2021 (UTC)

> I agree that some things are best adressed on a country-by-country level, as there are a lot of variation on the ground. However I fail to see how a source metadata field falls into that category.

Without the source, how will a mapper know/guess if a road geometry was added via a ground (accurate) or imagery (possible inaccurate) survey? This is critical to anyone doing ground surveys and improving maps locally.

> The same issue is present wherever you choose to put the source tag

I agree source=* should not be used but how about simply recommending using source:geometry and other detailed tags with a single value that represents the latest update?

> Just like mapping for the renderer, one shouldn't map for the editor ;)

Who should I map for then? Those making confusing wiki rules ? ;)

  • “As usual don't start deleting those tags indiscriminately unless and until there is a general project decision to do so.”
  • “The most common way to do this old-style source tagging is/was to”
  • “It is often preferable to use separate source tags for each of the main tags”
  • “However it seems the historic practice of tagging objects or individual attributes has not been officially deprecated yet, so don't go around deleting those source tags indiscriminately.”
"Without the source, how will a mapper know/guess if a road geometry was added via a ground (accurate) or imagery (possible inaccurate) survey? This is critical to anyone doing ground surveys and improving maps locally. " - look at the changesets that edited it. Note that typical road in well mapped areas was initially mapped from aerial, later edited based on other aerial data, with detail added via survey, more detail with aerial imagery data... It is unfeasible to explain and maintain such history in source=* Mateusz Konieczny (talk) 10:51, 19 August 2021 (UTC)
"how about simply recommending using source:geometry and other detailed tags with a single value that represents the latest update?" - almost always is ends not being updated and maintained so one needs to check edit history anyway Mateusz Konieczny (talk) 10:51, 19 August 2021 (UTC)
Thank you Mateusz Konieczny, I agree wholeheartedly. No need to repeat those arguments myself. (however I prefer associatedStreets relations, don't @ me :p) --Gileri (talk) 12:22, 21 August 2021 (UTC)

Plain URL

Mention how one can just say

source=https://example.org/nerdsome_files.html

or

source=https://www.freeway.example.gov/

Jidanni (talk) 04:38, 26 February 2023 (UTC)