Re: [Tagging] The current problem with tagging
On Mon, Oct 19, 2009 at 2:41 PM, Peter Childs wrote: > 2009/10/19 Martin Koppenhoefer : > The problem currently seams to be that there are 3 different places > within the wiki that things can be documented. Do you mean, Proposed/* pages, mapfeatures page and key:/tag: pages? > But if I want to map a Birthday Card Shop I could be here a while > trying to work out what the last person used Which may not be on > the wiki at all hence we end up with a mix of shop=card, > card=stationary, etc etc all of which may or may not be the same > thing Also you tend to read it every time as you don't remember > because you don't map them as often. And a birthday card shop is not the same as a postcard shop, but they can both be mapped as shop=card (or shop=gift according to mapfeatures). Some people will assume it's for birthday cards other thing it's for postcards.. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
On Mon, Oct 19, 2009 at 2:26 PM, Martin Koppenhoefer wrote: > 2009/10/19 Erik Johansson : > I'd say that 95% of the mappers are interested in what the > renderings show, and that is also in the mapfeatures-list. I agree > though that in the case you are looking for something special and not > already documented it will fail. Since I'm concentrating in the not documented tags that might explain why I dismiss it as irrelevant. But I really think it's now so big it is more of a hurdle than anything else, this is from watching newbies mapping on openstreetmap. >> Care to expand on what you mean? The wiki search sucks > > I don't agree that it sucks. My favorite feature are drinking But when you don't know the language and can't produce synonyms it certainly sucks, right? >> There is a need for personal documentation of tags, easy way to say >> what this tag means for you. IMO We are severely limited by English, >> what is easy to describe in Swedish for me, might be hard for me to do >> in German, Spanish or English. > > just go ahead. There is no limit, and everyone can document everything > on his personal wiki page (IMHO _this_ would lead to bad results, as > you get everyones own definition and are left alone for deciding, > which one you prefer, whilst today we maintain one page which displays > (at it's best) the consensous in definition. There must still be a standard, that is a consensus on a wikipage, there are several problems though: * You tag according to a standard, without understanding "You are doin it wrong" * You don't know there is a standard and tag as you want * There is no standard If you document what you think you are tagging then we have a better chance to understand what you mean by those two tiny key/val words you use to describe something. /Erik ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Tobias Knerr : > I just think that duplicating information that is already present in the > wiki or could easily be added to it (texts describing the meaning of > tags, images and some basic information such as the "Element" and > "Useful combination" sections from the default templates) should be avoided. It's a nice idea in theory, however storing it in 2 locations will make it harder to index/search the descriptions. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
John Smith: >> Which features of your sophisticated solution can NOT be properly >> implemented with a wiki backend? I don't see any except maybe the >> "voting" feature, and even if some data is best stored outside the wiki, >> that doesn't mean you cannot use the wiki for the rest. > > Can't easily keep the wiki updated from the number of uses either, or > order the returned information based on those pieces of information. Just to make this clear: I wasn't suggesting that you add those to the wiki. Instead, you could have your application aggregate documentation from the wiki, usage statistics calculated from OSM database, and some additional information only relevant for that application (such as vote counts) in order to present the user with an integrated, easily searchable view of everything related to tagging. I just think that duplicating information that is already present in the wiki or could easily be added to it (texts describing the meaning of tags, images and some basic information such as the "Element" and "Useful combination" sections from the default templates) should be avoided. Still, in the end it's all up to you. Tobias Knerr ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Martin Koppenhoefer : > 2009/10/19 Erik Johansson : >> On Mon, Oct 19, 2009 at 12:12 PM, Martin Koppenhoefer >> wrote: >>> 2009/10/19 John Smith : 2009/10/19 Martin Koppenhoefer : > 1. mapfeatures Only covers a very limited subset of tags. >>> >>> the most used ones. >> >> But that's not interesting what is needed is how to help people tag, >> and how to share tagging experiences between users. The most used ones >> are the least interesting. > > really? I'd say that 95% of the mappers are interested in what the > renderings show, and that is also in the mapfeatures-list. I agree > though that in the case you are looking for something special and not > already documented it will fail. > > 2. wiki-search Due to the wiki being used for everything this is very ineffective at times, do a search for lawyer... >>> >>> is an exception, there are not many exceptions though, and as soon as >>> a key:layer-page pops up, (e.g. set up by those who write continuously >>> about this lack in this list), it will be solved. >> >> Care to expand on what you mean? The wiki search sucks when you want >> to find out if someone else has documented a tag that describes what >> you want to do. > > I don't agree that it sucks. My favorite feature are drinking > fountains. I did a quick search on the wiki: > http://wiki.openstreetmap.org/wiki/Special:Search?search=potable+water&go=Go > > and the first result expanded to a nice page, that also mentioned > spring, fountain, water_point besides the obvious drinking_water. > Quite usable results IMHO. > >> There is a need for personal documentation of tags, easy way to say >> what this tag means for you. IMO We are severely limited by English, >> what is easy to describe in Swedish for me, might be hard for me to do >> in German, Spanish or English. > > just go ahead. There is no limit, and everyone can document everything > on his personal wiki page (IMHO _this_ would lead to bad results, as > you get everyones own definition and are left alone for deciding, > which one you prefer, whilst today we maintain one page which displays > (at it's best) the consensous in definition. > > cheers, > Martin > It seams to me that what we need is a better index rather than a new way of writing things down. The problem currently seams to be that there are 3 different places within the wiki that things can be documented. The regularly things are easy to find and also don't need looking at that often because the older mappers tend to remember them, Mostly people do not re-read the highways page, except when they first map one, But if I want to map a Birthday Card Shop I could be here a while trying to work out what the last person used Which may not be on the wiki at all hence we end up with a mix of shop=card, card=stationary, etc etc all of which may or may not be the same thing Also you tend to read it every time as you don't remember because you don't map them as often. Peter. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
> just go ahead. There is no limit, and everyone can document everything > on his personal wiki page (IMHO _this_ would lead to bad results, as > you get everyones own definition and are left alone for deciding, > which one you prefer, whilst today we maintain one page which displays > (at it's best) the consensous in definition. There's the issue of using the data at the back end some day. If someone documents a 'green spotted lemon shop' on his personal wiki page, what map data consumer is going to plow through personal wiki pages looking for results that might be relevant to his country / region? ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Erik Johansson : > On Mon, Oct 19, 2009 at 12:12 PM, Martin Koppenhoefer > wrote: >> 2009/10/19 John Smith : >>> 2009/10/19 Martin Koppenhoefer : 1. mapfeatures >>> >>> Only covers a very limited subset of tags. >> >> the most used ones. > > But that's not interesting what is needed is how to help people tag, > and how to share tagging experiences between users. The most used ones > are the least interesting. really? I'd say that 95% of the mappers are interested in what the renderings show, and that is also in the mapfeatures-list. I agree though that in the case you are looking for something special and not already documented it will fail. 2. wiki-search >>> >>> Due to the wiki being used for everything this is very ineffective at >>> times, do a search for lawyer... >> >> is an exception, there are not many exceptions though, and as soon as >> a key:layer-page pops up, (e.g. set up by those who write continuously >> about this lack in this list), it will be solved. > > Care to expand on what you mean? The wiki search sucks when you want > to find out if someone else has documented a tag that describes what > you want to do. I don't agree that it sucks. My favorite feature are drinking fountains. I did a quick search on the wiki: http://wiki.openstreetmap.org/wiki/Special:Search?search=potable+water&go=Go and the first result expanded to a nice page, that also mentioned spring, fountain, water_point besides the obvious drinking_water. Quite usable results IMHO. > There is a need for personal documentation of tags, easy way to say > what this tag means for you. IMO We are severely limited by English, > what is easy to describe in Swedish for me, might be hard for me to do > in German, Spanish or English. just go ahead. There is no limit, and everyone can document everything on his personal wiki page (IMHO _this_ would lead to bad results, as you get everyones own definition and are left alone for deciding, which one you prefer, whilst today we maintain one page which displays (at it's best) the consensous in definition. cheers, Martin ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Erik Johansson : > On Mon, Oct 19, 2009 at 1:55 PM, John Smith wrote: >> 2009/10/19 Tobias Knerr : >>> When I look at the problem, reimplementing all those features I've >>> mentioned is a lot more effort than writing a wiki text parser. >> >> The biggest problem is achieving consistency, how could you get that >> with the wiki, it's not designed to be machine parsible it's designed >> only to be human parsable... > > Who cares? Show us the tools that use this and how to use this data. > Sure there are lots of interesting things that could be done but you > have to show the gain for users and if they see this, I'm guessing, > they will gladly make things very machine parsable > > Don't ask other people to do work for you if you have nothing to give back. I guess you missed my earlier emails, I have been coding a proof of concept... http://tags.bigtincan.com/ ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
On Mon, Oct 19, 2009 at 1:55 PM, John Smith wrote: > 2009/10/19 Tobias Knerr : >> When I look at the problem, reimplementing all those features I've >> mentioned is a lot more effort than writing a wiki text parser. > > The biggest problem is achieving consistency, how could you get that > with the wiki, it's not designed to be machine parsible it's designed > only to be human parsable... Who cares? Show us the tools that use this and how to use this data. Sure there are lots of interesting things that could be done but you have to show the gain for users and if they see this, I'm guessing, they will gladly make things very machine parsable Don't ask other people to do work for you if you have nothing to give back. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
On Mon, Oct 19, 2009 at 12:12 PM, Martin Koppenhoefer wrote: > 2009/10/19 John Smith : >> 2009/10/19 Martin Koppenhoefer : >>> 1. mapfeatures >> >> Only covers a very limited subset of tags. > > the most used ones. But that's not interesting what is needed is how to help people tag, and how to share tagging experiences between users. The most used ones are the least interesting. >>> 2. wiki-search >> >> Due to the wiki being used for everything this is very ineffective at >> times, do a search for lawyer... > > is an exception, there are not many exceptions though, and as soon as > a key:layer-page pops up, (e.g. set up by those who write continuously > about this lack in this list), it will be solved. Care to expand on what you mean? The wiki search sucks when you want to find out if someone else has documented a tag that describes what you want to do. But this is also because people have too little time (or will) to consolidate pages into one, and the people that do this come off like megalomaniacs. There is a need for personal documentation of tags, easy way to say what this tag means for you. IMO We are severely limited by English, what is easy to describe in Swedish for me, might be hard for me to do in German, Spanish or English. -- /emj ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Tobias Knerr : > When I look at the problem, reimplementing all those features I've > mentioned is a lot more effort than writing a wiki text parser. The biggest problem is achieving consistency, how could you get that with the wiki, it's not designed to be machine parsible it's designed only to be human parsable... > If you offer a template, people tend to use it in a way that makes > sense. Of course you cannot strictly enforce it, but that doesn't mean > it wouldn't work. You still could use automated validation to spot for > potential problems if you think that's required. I think you are under estimating the problem, otherwise it would have already solved itself long ago, but by all means prove me wrong... > How does the backend used by a web app or editor feature determine > whether it's easy to use? Why is it harder to use a form whose results It doesn't, but being able to use it in an editor makes it easier to use and I don't see the current wiki being easy to update from within the editor... > are stored as a wiki template than a form whose results are written to > some database? Again, I don't see the wiki doing a good job of it, it does well what it was designed to do, making free text easy to make, it doesn't make it easy to use however. > That doesn't matter unless you manage to convince everyone to not use > the wiki at all but switch to your solution. If this doesn't work (which > isn't that unlikely, is it?) information has to be copied from and to > the wiki - suddenly the effort required is the added effort of both > solutions. Well feel free to make the wiki work, I don't think it's possible without completely re-writting large sections of code so it is useful, but I'd be happy to be shown how my opinion is wrong because then I wouldn't need to code anything to make something more useful than the wiki currently is. > Which features of your sophisticated solution can NOT be properly > implemented with a wiki backend? I don't see any except maybe the > "voting" feature, and even if some data is best stored outside the wiki, > that doesn't mean you cannot use the wiki for the rest. Can't easily keep the wiki updated from the number of uses either, or order the returned information based on those pieces of information. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
John Smith: > 2009/10/19 Tobias Knerr : >> I'd strongly recommend to use the >> wiki as your backend for most of the information. > > It doesn't matter how I look at the problem there will need to be a > lot of effort exerted to be useful. When I look at the problem, reimplementing all those features I've mentioned is a lot more effort than writing a wiki text parser. > The users need to spend a lot of time/effort getting the wiki up to > scratch, even then there is no guarantee that will be good enough > because there is no code at present to enforce some sort of > consistency needed to build a useful/encompassing index. If you offer a template, people tend to use it in a way that makes sense. Of course you cannot strictly enforce it, but that doesn't mean it wouldn't work. You still could use automated validation to spot for potential problems if you think that's required. > Which is the > point of this thread to make is as simple for people to use and keep > up to date as possible with the least amount of effort as possible. How does the backend used by a web app or editor feature determine whether it's easy to use? Why is it harder to use a form whose results are stored as a wiki template than a form whose results are written to some database? > You can't do that with a wiki, and as I see it the amount of effort to > get the wiki to contort into something useful is about the same as > doing something new, although the learning curve to integrate with the > wiki might actually make it more effort to use a wiki for something as > sophisticated as what I'm proposing. That doesn't matter unless you manage to convince everyone to not use the wiki at all but switch to your solution. If this doesn't work (which isn't that unlikely, is it?) information has to be copied from and to the wiki - suddenly the effort required is the added effort of both solutions. Which features of your sophisticated solution can NOT be properly implemented with a wiki backend? I don't see any except maybe the "voting" feature, and even if some data is best stored outside the wiki, that doesn't mean you cannot use the wiki for the rest. Tobias Knerr ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Tobias Knerr : > Whether you try to continue that Semantic MediaWiki approach that has > been suggested some time ago > http://wiki.openstreetmap.org/wiki/Semantic_MediaWiki > or create your own app for browsing/searching (probably using the API > (http://wiki.openstreetmap.org/api.php) together with some parsing to > extract templates) is up to you, but I'd strongly recommend to use the > wiki as your backend for most of the information. It doesn't matter how I look at the problem there will need to be a lot of effort exerted to be useful. The users need to spend a lot of time/effort getting the wiki up to scratch, even then there is no guarantee that will be good enough because there is no code at present to enforce some sort of consistency needed to build a useful/encompassing index. Which is the point of this thread to make is as simple for people to use and keep up to date as possible with the least amount of effort as possible. You can't do that with a wiki, and as I see it the amount of effort to get the wiki to contort into something useful is about the same as doing something new, although the learning curve to integrate with the wiki might actually make it more effort to use a wiki for something as sophisticated as what I'm proposing. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
John Smith wrote:
> As an example I added the 'highway' tag, and 'motorway' value and
> copied the descriptions from the wiki.
>
> The list of things that would need to be done to make this actually
> useful include:
> [...]
* Avoid inconsistencies with existing documentation (that is: wiki)
Imo that's one of the most important features. And it's best done by
simply using the wiki for storing that information and extracting the
relevant parts in a machine-readable way.
Also, the wiki already offers these:
> * edit/update tags and values
> * extend the wiki formatting function to include more than just bold (''')
> * add example images, or reference URLs so these could optionally be
> displayed by editors/web site
> * the ability to add context to tags by linking tags that a commonly
> used in conjunction with each other, eg lanes=* should be linked to
> highway=*
as well as
* watchlists and notification
* documented API for content access, existing bot/automation tools
* history, edits, edit comments
* user accounts, permission management
* ...
Imo, you should try to figure out what information is missing from the
wiki and add it there (probably using templates) instead of creating a
completely independent documentation database.
Whether you try to continue that Semantic MediaWiki approach that has
been suggested some time ago
http://wiki.openstreetmap.org/wiki/Semantic_MediaWiki
or create your own app for browsing/searching (probably using the API
(http://wiki.openstreetmap.org/api.php) together with some parsing to
extract templates) is up to you, but I'd strongly recommend to use the
wiki as your backend for most of the information.
Tobias Knerr
___
Tagging mailing list
Tagging@openstreetmap.org
http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Martin Koppenhoefer : > is an exception, there are not many exceptions though, and as soon as > a key:layer-page pops up, (e.g. set up by those who write continuously > about this lack in this list), it will be solved. This requires an out of band work process, that is stop mapping and add something to the wiki or spend a lot of time doing this to find out someone already made a similar tag, I don't see this as a very good idea. >> This has shown to be very ineffective with contensious tags or it ends >> up with personal attacks because some think there is a predefined >> limit of posts people can make... > > in the end you didn't care ;-) I do care about the contentious issues, they seem to be a complete waste of time and a better system needs to be developed. As for the other, that is a subjective thing and I won't rebother rehashing arguments on this as it's completely off topic. I was using it to highlight my point that using mailing lists according to some isn't an option. > the search is not that bad. Besides some exceptions (with too many > results) you usually find the stuff you're looking for (mostly if it > is about keys). The search has a number of failings which I've already pointed out either in emails or in the desired feature list I think would be useful. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Martin Koppenhoefer : > In case that you have to invent a new key, it is obvious that more > work is involved, in order to make it good. Which is what I'm trying to figure out :) (and why I started this thread) The current system is widely criticised, but no one has suggested a better option. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 John Smith : > 2009/10/19 Martin Koppenhoefer : >> 1. mapfeatures > > Only covers a very limited subset of tags. the most used ones. >> 2. wiki-search > > Due to the wiki being used for everything this is very ineffective at > times, do a search for lawyer... is an exception, there are not many exceptions though, and as soon as a key:layer-page pops up, (e.g. set up by those who write continuously about this lack in this list), it will be solved. >> 3. (OSM-Doc / tagwatch) > > Only gives a single key pair, doesn't highlight combinations or > context tags are used with each other. true, could be augmented >> 0./4. mailing-list / forum > > This has shown to be very ineffective with contensious tags or it ends > up with personal attacks because some think there is a predefined > limit of posts people can make... in the end you didn't care ;-) >> 5. own tag (preferably plus documentation) > While you may be able to document something, it's difficult for people > to find it after wards, which is the whole point I presented from my > original email, including why the wiki/tag watch don't scale. the search is not that bad. Besides some exceptions (with too many results) you usually find the stuff you're looking for (mostly if it is about keys). cheers, Martin ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Liz : > On Mon, 19 Oct 2009, Martin Koppenhoefer wrote: >> 2009/10/19 John Smith : >> > That may be the case, but doesn't it make sense to make it as easy as >> > possible for people to find tags already in use? and use one of those >> > before making a new tag that does the same thing? >> >> 1. mapfeatures >> 2. wiki-search >> 3. (OSM-Doc / tagwatch) >> 0./4. mailing-list / forum >> 5. own tag (preferably plus documentation) >> >> cheers, >> Martin >> > So that is 5 steps. > Maybe 7 > Do you see that some sort of streamlining would be worthwhile? no. Because it is 5 Steps for features that are new, that is, nobody ever documented or even used. Usually you will have 1 step: mapfeatures. E.g. a motorway: you got there, take the key, read about useful combinations, and map. In case that you have to invent a new key, it is obvious that more work is involved, in order to make it good. cheers, Martin ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Martin Koppenhoefer : > 1. mapfeatures Only covers a very limited subset of tags. > 2. wiki-search Due to the wiki being used for everything this is very ineffective at times, do a search for lawyer... > 3. (OSM-Doc / tagwatch) Only gives a single key pair, doesn't highlight combinations or context tags are used with each other. > 0./4. mailing-list / forum This has shown to be very ineffective with contensious tags or it ends up with personal attacks because some think there is a predefined limit of posts people can make... > 5. own tag (preferably plus documentation) While you may be able to document something, it's difficult for people to find it after wards, which is the whole point I presented from my original email, including why the wiki/tag watch don't scale. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
On Mon, 19 Oct 2009, Martin Koppenhoefer wrote: > 2009/10/19 John Smith : > > That may be the case, but doesn't it make sense to make it as easy as > > possible for people to find tags already in use? and use one of those > > before making a new tag that does the same thing? > > 1. mapfeatures > 2. wiki-search > 3. (OSM-Doc / tagwatch) > 0./4. mailing-list / forum > 5. own tag (preferably plus documentation) > > cheers, > Martin > So that is 5 steps. Maybe 7 Do you see that some sort of streamlining would be worthwhile? Liz ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 John Smith : > That may be the case, but doesn't it make sense to make it as easy as > possible for people to find tags already in use? and use one of those > before making a new tag that does the same thing? 1. mapfeatures 2. wiki-search 3. (OSM-Doc / tagwatch) 0./4. mailing-list / forum 5. own tag (preferably plus documentation) cheers, Martin ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Martin Koppenhoefer : > 2009/10/18 Liz : >> As a new mapper, about 2 years ago, I thought that the only POIs OSM >> collected >> were pubs post offices, banks and other things that were in the editor. >> I didn't get the slightest hint from the wiki front page that there were >> options for "anything you find". > > It is the first phrase on the mapfeatures-page though (the page I > always went when I was new to OSM and was looking for a tag): > "OpenStreetMap does not have any content restrictions on tags that can > be assigned to OSM-Elements (Nodes Node, Ways Way or Relations > Relation). You can use any tags you like as long as the values are > verifiable. " That may be the case, but doesn't it make sense to make it as easy as possible for people to find tags already in use? and use one of those before making a new tag that does the same thing? ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/18 Liz : > As a new mapper, about 2 years ago, I thought that the only POIs OSM collected > were pubs post offices, banks and other things that were in the editor. > I didn't get the slightest hint from the wiki front page that there were > options for "anything you find". It is the first phrase on the mapfeatures-page though (the page I always went when I was new to OSM and was looking for a tag): "OpenStreetMap does not have any content restrictions on tags that can be assigned to OSM-Elements (Nodes Node, Ways Way or Relations Relation). You can use any tags you like as long as the values are verifiable. " cheers, Martin ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
I wrote up a wiki page, including a possible feature set. http://wiki.openstreetmap.org/wiki/Machine_Readable_Tagging_API ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
I've coded up a VERY VERY basic proof of concept:
http://tags.bigtincan.com
So far you can search for a specific tag name, view the description of
the tag, add a tag and add values for a tag and a description of the
value.
As an example I added the 'highway' tag, and 'motorway' value and
copied the descriptions from the wiki.
The list of things that would need to be done to make this actually
useful include:
* allow tags and values to be easily translated.
* edit/update tags and values
* allow people to vote the weighting of tags up/down
* allow people to add/remove synonyms for tags/values, and be able to
search on the synonyms
* extend the wiki formatting function to include more than just bold (''')
* extend the search to do fuzzy searching of values not just tags
* write an API so editors can hook into it to read data and write data
back into the system
* needs some kind of caching mechanism so editors only need to download updates
* add example images, or reference URLs so these could optionally be
displayed by editors/web site
* the ability to add context to tags by linking tags that a commonly
used in conjunction with each other, eg lanes=* should be linked to
highway=*
I'm sure there is more, please let me know if you think of things, but
that's all I've been able to come up with so far.
___
Tagging mailing list
Tagging@openstreetmap.org
http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/19 Mike N. : > The case I run into is "How do I tag y" - the wiki is well laid out, but > the answer is often not as accessible as it should be. I tried to tag law > offices in a meaningful way. Searching for Lawyer, Legal, Barrister even in > the proposed pages brought up nothing. I can't believe I'm the first person > to want to do this. I should have used Google because there is a tag called > shop=lawyer on the 'Abandoned' Wiki pages. That's a good idea as well, instead of just searching on the name of a tag and the description we should also store synonyms as well so they can be searched on as well. Tagging tags :) ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
> Searching the wiki is tedious, at least someone else realised that a > formal > search engine did it more efficiently, and has let me know. http://www.google.com/#hl=en&source=hp&q=insurance+site%3Awiki.openstreetmap.org&aq=f&aqi=&oq=&fp=c807e9ccc08a197a In this case, it found the relevant result, along with many banned spam pages. Other results could vary. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
On Mon, 19 Oct 2009, Tobias Knerr wrote: > Mike N.: > > The case I run into is "How do I tag y" - the wiki is well laid out, > > but the answer is often not as accessible as it should be. I tried to > > tag law offices in a meaningful way. Searching for Lawyer, Legal, > > Barrister even in the proposed pages brought up nothing. I can't believe > > I'm the first person to want to do this. I should have used Google > > because there is a tag called shop=lawyer on the 'Abandoned' Wiki pages. > > Why google? There is a search box in the wiki that I'd use first: > http://wiki.openstreetmap.org/index.php?title=Special:Search&search=lawyer > > First hit is the abandoned proposal amenity=lawyer. Unfortunately, > there's a lot of ODBL and licensing results, too, so maybe we could > introduce a namespace for tagging so you could only search tag > documentation and proposals? > > > The point is that scholarly research is time consuming, and most people > > don't have the patience and we're possibly discouraging mappers who would > > otherwise contribute. > > Do you have an easier solution than a search box? Integration of that > search box with editors would be nice (these could then search their > presets, too), but I don't think there is a more useful user interface > for answering the "How do I tag y" question. > > By the way: We also have a German page for "natural language -> tags" in > the wiki (http://wiki.openstreetmap.org/wiki/DE:Howto_Map_A) which is > maintained manually. The English equivalent is rather limited, though. > > Tobias Knerr > Tobias, the wiki search is extremely frustating. I will have multiple main wiki pages open the abandoned, the voted out, the maybes and the accepted. I search them all and then have to find out why something was abandoned. Often things are abandoned because of a time interval, but have not been superseded by anything. This eventually produces a lawyer tag on the search system. That takes about 15 minutes to complete and decide on what to do. That's why the machine readable tag is being suggested, translations done once at the API level and able to be picked up by editors automatically. As a new mapper, about 2 years ago, I thought that the only POIs OSM collected were pubs post offices, banks and other things that were in the editor. I didn't get the slightest hint from the wiki front page that there were options for "anything you find". Searching the wiki is tedious, at least someone else realised that a formal search engine did it more efficiently, and has let me know. Liz ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
My main points from my original post still stands, I feel that the wiki is inadequate to deal with the task of maintaining an index of tag definitions properly, it is simply too much effort to be useful for less common tags because you have to spend too much time searching which takes time away from mapping. If the process could be made easier, or better yet, part of the work flow I think things would be a lot more productive then they are now. Another weighting system I thought of after my original post could be a voting system where people vote tags up or down, but instead of including or excluding this only effects the order that results are displayed from a search. I also think if an API is found to be useful mapping editors will adopt this themselves because it will make life easier for their users. Currently I code in PHP and java, so I'm not sure how much help I could be in coding a solution, but I am willing to spend time making a proof of concept. Has anyone coded something similar to this already? Specifically the translation side of things, because no only does the tag name need to be translated but some key names would also need to be translated as well. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
Hi! Peteris Krisjanis schrieb: >> The discussions over the last few months indicate a lot of people feel >> there needs to be a more formal tag approval process. > > I think not a lot, but majority. Most of us want consistent map data, > even that "another camp". +1 > However, I think there is nor problem nor conflict at all - simply > people who cares for tagging works on that and outputs results like > proposed and accepted features, etc. I think other mappers will use > these tags sooner or later. Introducing new tags and tagging schemes for new features is not the problem - that works fairly well. The problem are those tags that are in widespread use, but are ambiguously defined. Where the same tag is used with different meanings by different schools of mappers, we have the real problem. If we'd introduce some advanced feature/tag finding system, we would still have to decide on a single definition for every tag offered there. Discussion has brought us nowhere, so someone needs to make a hard decision which interpretation to include in the lookup system and which to drop. bye Nop ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
On 18/10/2009 17:30, Peteris Krisjanis wrote: > What we need is - this list where all people who cares about tagging > (and it already shows that most tagging decissions will be made in > this part of community) and some kind of wiki book 'OSM Mapping for > dummies' where could be collected know-hows and howtos (how example to > tag crossroads) in good editorial quality. I'm not disagreeing with everything else you said. My main aim though is to try to make the documentation and tag schema machine readable, so that it is useful to more than just human readers. The wiki page or future equivalent would be derived from the schema, but so would the presets in the editors (in principle - we can't mandate this, obviously) and the checking criteria in error checkers among other things (yes, I know there's more to error checking than just the tag names and values) and some of the rendering updates when tags change. David ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
> Why google? There is a search box in the wiki that I'd use first: > http://wiki.openstreetmap.org/index.php?title=Special:Search&search=lawyer > > First hit is the abandoned proposal amenity=lawyer. Unfortunately, > there's a lot of ODBL and licensing results, too, so maybe we could > introduce a namespace for tagging so you could only search tag > documentation and proposals? Being able to restrict a namespace to tagging would be a big help. Try insurance (for a shop which sells insurance). I could swear that I created an entry on the proposed shops page for insurance, but a wiki search didn't find it. > Do you have an easier solution than a search box? Integration of that > search box with editors would be nice (these could then search their > presets, too), but I don't think there is a more useful user interface > for answering the "How do I tag y" question. It would be good to add to the editors - the law of the internet is that search boxes on web sites are worse than useless. A search for a street address on the main maps page only reinforces this.If we can organize the wiki in to facilitate useful search functions that can be integrated into editors, that would solve most of the issues. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
On 18/10/2009 15:59, Anthony wrote: > On Sun, Oct 18, 2009 at 10:08 AM, David Earl > wrote: >> Which is why I am seeing if we can find a middle way: one that makes >> tags more accessible and manageable in automated fashion, but doesn't >> limit people's freedom to innovate and suggest by example. > > A "middle way" is usually either 1) a "solution" where no one is > happy; or 2) a temporary step along a slippery slope from A to B, > which is implemented to get from A to B without the proponents of A > complaining too loudly. > > Let's hope this one winds up being 2 and not 1 :). A middle way only in that it prevents neither way of working, not in the sense of a compromise that doesn't satisfy either camp. No, I hope it provides advantages for everyone: by having centralised machine readable tag descriptions, all the applications can work off the same schema, we can avoid typos and/or have common error checking, while still being able to add tags, and it allows them to be advertised and documented so there's a better chance of them being adopted rather than reinvented. We can make links between them so similar or identical or superseded tags can be flagged. We can describe them in multiple languages in one place accessible to everyone, not just human readers of a web page (or screen scrapers). We can have synonyms to allow for multiple languages. We can indicate what "property" tags are considered appropriate to what "object" tags. And so on. Basically I think we can do more, and more robustly, without squashing intrinsic freedoms that are there now. David. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/18 David Earl : > On 18/10/2009 14:53, Anthony wrote: >> On Sun, Oct 18, 2009 at 8:32 AM, David Earl >> wrote: >>> On 18/10/2009 12:01, John Smith wrote: That's a good point, having some kind of API for tags would make it possible to have translations with very little extra effort from a programming point of view. >>> Indeed. See my message from earlier this month: >>> http://lists.openstreetmap.org/pipermail/talk/2009-October/043340.html >> >> I'd go further than that, and only allow tags marked as "experimental" >> (for instance by prepending "X-") to be used without an open >> discussion and approval by a team of experts. There are some really >> ugly inconsistencies in the OSM tagging scheme, and without expert >> help these inconsistencies are going to get worse and worse. > > The discussions over the last few months indicate a lot of people feel > there needs to be a more formal tag approval process. I think not a lot, but majority. Most of us want consistent map data, even that "another camp". > The problem is there is an equally strong feeling from another camp of > people who think exactly the opposite, that tagging should be a > completely free activity which converges on solutions by consensus and > popularity. (I get the impression most such contributors have stopped > even trying to take part in discussions, and just get on with it. Many > won't have subscribed to this list). As far as I see, problem is not that they don't want to have tags already defined. They just don't want to get themselves to be dragged into very long acceptance process. And they have a point - serious mapping can be very time consuming already and when your free time is limited, you vote for short-term gain - what is simply mapping fun. However, I think there is nor problem nor conflict at all - simply people who cares for tagging works on that and outputs results like proposed and accepted features, etc. I think other mappers will use these tags sooner or later. > Which is why I am seeing if we can find a middle way: one that makes > tags more accessible and manageable in automated fashion, but doesn't > limit people's freedom to innovate and suggest by example. No, that would be hard to achieve and in fact no one wants that. I really doubt that 'mappers for fun' really want freedom to tag as they want - they want their data to be usable, don't they? They just don't want to think about it. If there's someone already done all thinking and voting process, why not use tags already offered by them? They want to a) map it b) throw data in OSM c) tag it and be done. What we need is - this list where all people who cares about tagging (and it already shows that most tagging decissions will be made in this part of community) and some kind of wiki book 'OSM Mapping for dummies' where could be collected know-hows and howtos (how example to tag crossroads) in good editorial quality. Cheers, Peter. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
Mike N.: > The case I run into is "How do I tag y" - the wiki is well laid out, but > the answer is often not as accessible as it should be. I tried to tag law > offices in a meaningful way. Searching for Lawyer, Legal, Barrister even in > the proposed pages brought up nothing. I can't believe I'm the first person > to want to do this. I should have used Google because there is a tag called > shop=lawyer on the 'Abandoned' Wiki pages. Why google? There is a search box in the wiki that I'd use first: http://wiki.openstreetmap.org/index.php?title=Special:Search&search=lawyer First hit is the abandoned proposal amenity=lawyer. Unfortunately, there's a lot of ODBL and licensing results, too, so maybe we could introduce a namespace for tagging so you could only search tag documentation and proposals? > The point is that scholarly research is time consuming, and most people > don't have the patience and we're possibly discouraging mappers who would > otherwise contribute. Do you have an easier solution than a search box? Integration of that search box with editors would be nice (these could then search their presets, too), but I don't think there is a more useful user interface for answering the "How do I tag y" question. By the way: We also have a German page for "natural language -> tags" in the wiki (http://wiki.openstreetmap.org/wiki/DE:Howto_Map_A) which is maintained manually. The English equivalent is rather limited, though. Tobias Knerr ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
>> the wiki allows >> people to easily document things, but you can't easily find them and >> I'm talking about everyone here, especially those that are only likely >> to use the map features page. > > Can you give an example here? I actually don't know what situation you > are thinking about. Answering "what does tag x mean" is easy. "How do I > tag y" can usually be answered using the wiki search (or browser search > on Map Features). Tags that can used in addition to a given feature/tag > are listed, too, and it's not that hard to just manually add a new tag > there. So what's missing here? The case I run into is "How do I tag y" - the wiki is well laid out, but the answer is often not as accessible as it should be. I tried to tag law offices in a meaningful way. Searching for Lawyer, Legal, Barrister even in the proposed pages brought up nothing. I can't believe I'm the first person to want to do this. I should have used Google because there is a tag called shop=lawyer on the 'Abandoned' Wiki pages. The point is that scholarly research is time consuming, and most people don't have the patience and we're possibly discouraging mappers who would otherwise contribute. It also promotes tagging with the first thing that pops into the mind, which may be the wrong tag ( drinking fountain -> fountain = wrong ). The JOSM preset "More information about this feature" has saved me from tagging things wrongly a number of times. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
On Sun, Oct 18, 2009 at 10:08 AM, David Earl wrote: > Which is why I am seeing if we can find a middle way: one that makes > tags more accessible and manageable in automated fashion, but doesn't > limit people's freedom to innovate and suggest by example. A "middle way" is usually either 1) a "solution" where no one is happy; or 2) a temporary step along a slippery slope from A to B, which is implemented to get from A to B without the proponents of A complaining too loudly. Let's hope this one winds up being 2 and not 1 :). ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
On 18/10/2009 14:53, Anthony wrote: > On Sun, Oct 18, 2009 at 8:32 AM, David Earl > wrote: >> On 18/10/2009 12:01, John Smith wrote: >>> That's a good point, having some kind of API for tags would make it >>> possible to have translations with very little extra effort from a >>> programming point of view. >> Indeed. See my message from earlier this month: >> http://lists.openstreetmap.org/pipermail/talk/2009-October/043340.html > > I'd go further than that, and only allow tags marked as "experimental" > (for instance by prepending "X-") to be used without an open > discussion and approval by a team of experts. There are some really > ugly inconsistencies in the OSM tagging scheme, and without expert > help these inconsistencies are going to get worse and worse. The discussions over the last few months indicate a lot of people feel there needs to be a more formal tag approval process. The problem is there is an equally strong feeling from another camp of people who think exactly the opposite, that tagging should be a completely free activity which converges on solutions by consensus and popularity. (I get the impression most such contributors have stopped even trying to take part in discussions, and just get on with it. Many won't have subscribed to this list). Which is why I am seeing if we can find a middle way: one that makes tags more accessible and manageable in automated fashion, but doesn't limit people's freedom to innovate and suggest by example. David ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
On Sun, Oct 18, 2009 at 8:32 AM, David Earl wrote: > On 18/10/2009 12:01, John Smith wrote: >> That's a good point, having some kind of API for tags would make it >> possible to have translations with very little extra effort from a >> programming point of view. > > Indeed. See my message from earlier this month: > http://lists.openstreetmap.org/pipermail/talk/2009-October/043340.html I'd go further than that, and only allow tags marked as "experimental" (for instance by prepending "X-") to be used without an open discussion and approval by a team of experts. There are some really ugly inconsistencies in the OSM tagging scheme, and without expert help these inconsistencies are going to get worse and worse. And that only gets us to the point where we can tackle the bigger and more difficult problem. OSM is doing something that no one in the world has ever really done - combining all the open content geographical data ever produced into "one GIS to rule them all". ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
John Smith: > the wiki allows > people to easily document things, but you can't easily find them and > I'm talking about everyone here, especially those that are only likely > to use the map features page. Can you give an example here? I actually don't know what situation you are thinking about. Answering "what does tag x mean" is easy. "How do I tag y" can usually be answered using the wiki search (or browser search on Map Features). Tags that can used in addition to a given feature/tag are listed, too, and it's not that hard to just manually add a new tag there. So what's missing here? >>From there if an API could be created for editors we could do things > like mousing over a tag in JOSM/potlatch then throws up a page of text > describing the tag. Why do we need an API for this? The wiki pages already have standardized names and standardized templates, so it should be possible to extract information from there. > Any tags the are entered that aren't in the database could generate a > prompt asking people to document it on the spot, Considering that ambiguous tag definitions are probably the worst tagging problem we have to deal with in the long run, I'm not sure whether the definitions created this way wouldn't cause more problems than they would solve. Tobias Knerr ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
On 18/10/2009 12:01, John Smith wrote: > 2009/10/18 Mike N. : >>> From there if an API could be created for editors we could do things >>> like mousing over a tag in JOSM/potlatch then throws up a page of text >>> describing the tag. >>> >>> Any tags the are entered that aren't in the database could generate a >>> prompt asking people to document it on the spot, alternatively we >>> could have a suggestion engine by searching both tag names and in the >>> descriptions and seeing if something similar already exists and they >>> could choose to use that instead. >> I agree with this - someone else described a system of meta-tags for tags, >> integrated into the editors to make it easy to locate the tag you're >> thinking of by subject, possibly even in the local language. The editors >> could even hide the 'official English tag' used internally by OSM. Less >> common local items could be in a language other than English. > > That's a good point, having some kind of API for tags would make it > possible to have translations with very little extra effort from a > programming point of view. Indeed. See my message from earlier this month: http://lists.openstreetmap.org/pipermail/talk/2009-October/043340.html David ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
2009/10/18 Mike N. : >> From there if an API could be created for editors we could do things >> like mousing over a tag in JOSM/potlatch then throws up a page of text >> describing the tag. >> >> Any tags the are entered that aren't in the database could generate a >> prompt asking people to document it on the spot, alternatively we >> could have a suggestion engine by searching both tag names and in the >> descriptions and seeing if something similar already exists and they >> could choose to use that instead. > > I agree with this - someone else described a system of meta-tags for tags, > integrated into the editors to make it easy to locate the tag you're > thinking of by subject, possibly even in the local language. The editors > could even hide the 'official English tag' used internally by OSM. Less > common local items could be in a language other than English. That's a good point, having some kind of API for tags would make it possible to have translations with very little extra effort from a programming point of view. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] The current problem with tagging
> From there if an API could be created for editors we could do things > like mousing over a tag in JOSM/potlatch then throws up a page of text > describing the tag. > > Any tags the are entered that aren't in the database could generate a > prompt asking people to document it on the spot, alternatively we > could have a suggestion engine by searching both tag names and in the > descriptions and seeing if something similar already exists and they > could choose to use that instead. I agree with this - someone else described a system of meta-tags for tags, integrated into the editors to make it easy to locate the tag you're thinking of by subject, possibly even in the local language. The editors could even hide the 'official English tag' used internally by OSM. Less common local items could be in a language other than English. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
[Tagging] The current problem with tagging
I've been mulling this over in the last few days, but it appears to me the biggest issue at present is one of documentation, specifically putting things into context and more importantly making them easily found. So far nothing seem to fulfill the above criteria, the wiki allows people to easily document things, but you can't easily find them and I'm talking about everyone here, especially those that are only likely to use the map features page. On the flip side, things like tagwatch make is sort of easy to find tags in use but it doesn't give you context needed to work out not only what is being used, but what other tags are they used in conjuction with. I don't think it's helpful to tell people to just use tags or document tags you are using, because you can't correlate this information with what everyone else is doing, and that is where some attention needs to be focused if the data going in is going to be useful at all coming out again. I don't have a solution to this problem, at least not a full solution, but I have come up with the following that might push things closer to a solution: We need something a little more capable than a wiki, or we need some kind of front end to collate the information in the wiki into a useable index. The data from the index needs to cross reference with the OSM DB data, so that we can not only know what other people are doing, especially those within our own regions, but to then make it easy for people to document what those tags are being used for. >From there if an API could be created for editors we could do things like mousing over a tag in JOSM/potlatch then throws up a page of text describing the tag. Any tags the are entered that aren't in the database could generate a prompt asking people to document it on the spot, alternatively we could have a suggestion engine by searching both tag names and in the descriptions and seeing if something similar already exists and they could choose to use that instead. Since we are dealing with geospatial information we could even weight things based on the number of times a tag is used within say 100-1000km. That's all I've come up with so far. ___ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
