Re: [HOT] Buildings in Chad
bounding box in JOSM lat 10.8470757 10.8482862 long 15.0446431 15.0463866 There are several. I'll leave them for you to experiment with. They are in chad2.osm or pick them up here https://www.jatws.org/johnw/buildingexample.zip Cheerio John On 16 July 2017 at 15:27, Bjoern Hassler wrote: > Hi John, > > Could you send me example coordinates or task number? > > Bjoern > > > On 15 Jul 2017 16:37, "john whelan" wrote: > > I suspect your standards are much higher than mine. Basically I ignore if > a building is square or not, in general I ignore buildings unless they have > been mapped two or three times in which case if they catch my eye I will > delete some of the multiple copies. It might be worthwhile building a > script to pick these out I see quite a few of them but you'd need to set up > a team to clean them up. > > If a settlement has been tagged a building then if I notice it I'll retag > it landuse=residential. A script might be useful for picking these out, > pick out those buildings that are above a predefined size. Again you need > someone to inspect them. > > On buildings I look purely for untagged ways and whilst the todo plugin > can be useful sometimes just looking for the untagged by mapper will give > you a cluster of untagged buildings that you can tag all at once. The todo > plugin is used on the last pass. Mixed in with the untagged buildings will > be highways and other features. > > I have noticed that individual mappers can be very productive. Especially > after feedback, whilst maperthons are fine they take up travel time etc. so > old aged pensioners or retirees often are less well represented and once > they get the hang of it they have the time and patience to map. > > The other side of maperthons is it takes a fair bit of organising to do > one properly and anyone can hold one. From the names of the mappers and > the time of day when they map it is doubtful if many mappers first language > is English. What I'm looking at is the map after everyone has mapped and > the mappers do have different skill levels and training. > > So for a well organised mapathon your tools will be useful. For the rest > of the world I suspect it will be up to other mappers to clean up slowly. > > Cheerio John > > On 15 July 2017 at 08:51, Bjoern Hassler wrote: > >> Hi John, >> >> At the recent Cambridge mapathon, we used this tool >> https://github.com/OSM-Utilities/JOSM-Scripts-HOT/tree/master/www >> (which we developed especially for mapathons). >> >> You enter the hastags and time of your mapathon, and it produces a list >> of changes made by those users during the mapathon, that is "validated" >> against certain criteria, e.g. whether 4-node buildings are square, whether >> there are untagged isolated nodes, etc. The tool gives you a percentage: >> >> - User ABC: Non-square buildings 2% >> ... >> - User DEF: Non-square buildings 95% >> ... >> >> So it means that during the mapathon, you can see who just makes >> occasional slips vs. systematic mistakes. >> >> At the Cambridge mapathon we used this quite successfully, seeing which >> users were putting in a lot of buildings that were not squared (~ 100s of >> buildings in two cases). We managed to find them (which of course is not >> necessarily trivial), and they corrected the issue before the mapathon was >> over. This fixes both the buildings they added (one-off as it were), but >> also means they will square buildings in future. >> >> Another use of the tool is to identify where a particular mapathon as a >> whole generates lots of non-square buildings, in which case the supporters >> need a bit of additional support :) Also, you can tell who doesn't map much >> during the mapathon: They might need a bit more encouragement, to make sure >> they don't get frustrated, and to make sure they come back. >> >> Even though I say this myself, I think the tool has great promise, and we >> demo'd this at the last London mid-month mapathon, so some people will have >> seen it. We're also doing a webinar on this later this month. The tool is >> really just mashup of the RedCross hashtag API and overpass, but I haven't >> seen any other tools working like this. (E.g. in osmcha, you can see issues >> with changesets, which is great, but afaik it doesn't allow me to see >> whether users are make occasional slips, vs. systematic errors.) >> >> Unfortunately the RedCross api server does not support (see https: >> https://github.com/AmericanRedCross/osm-stats/issues/47). Thus we cannot >> have an online version, but you need to pull the repo and run locally. >> Hopefully this will come very soon. >> >> The 2nd issue is that it can be hard to find users. I wished that the >> tasks manager had a way to register users who are doing a mapathon, i.e. as >> you start, you put a table number in, which means helpers can find you on >> the basis of your OSM id. >> >> For your use case, it would be possible to get the script to produce a >> list of non-
Re: [HOT] Buildings in Chad
Hi John, Could you send me example coordinates or task number? Bjoern On 15 Jul 2017 16:37, "john whelan" wrote: I suspect your standards are much higher than mine. Basically I ignore if a building is square or not, in general I ignore buildings unless they have been mapped two or three times in which case if they catch my eye I will delete some of the multiple copies. It might be worthwhile building a script to pick these out I see quite a few of them but you'd need to set up a team to clean them up. If a settlement has been tagged a building then if I notice it I'll retag it landuse=residential. A script might be useful for picking these out, pick out those buildings that are above a predefined size. Again you need someone to inspect them. On buildings I look purely for untagged ways and whilst the todo plugin can be useful sometimes just looking for the untagged by mapper will give you a cluster of untagged buildings that you can tag all at once. The todo plugin is used on the last pass. Mixed in with the untagged buildings will be highways and other features. I have noticed that individual mappers can be very productive. Especially after feedback, whilst maperthons are fine they take up travel time etc. so old aged pensioners or retirees often are less well represented and once they get the hang of it they have the time and patience to map. The other side of maperthons is it takes a fair bit of organising to do one properly and anyone can hold one. From the names of the mappers and the time of day when they map it is doubtful if many mappers first language is English. What I'm looking at is the map after everyone has mapped and the mappers do have different skill levels and training. So for a well organised mapathon your tools will be useful. For the rest of the world I suspect it will be up to other mappers to clean up slowly. Cheerio John On 15 July 2017 at 08:51, Bjoern Hassler wrote: > Hi John, > > At the recent Cambridge mapathon, we used this tool > https://github.com/OSM-Utilities/JOSM-Scripts-HOT/tree/master/www > (which we developed especially for mapathons). > > You enter the hastags and time of your mapathon, and it produces a list of > changes made by those users during the mapathon, that is "validated" > against certain criteria, e.g. whether 4-node buildings are square, whether > there are untagged isolated nodes, etc. The tool gives you a percentage: > > - User ABC: Non-square buildings 2% > ... > - User DEF: Non-square buildings 95% > ... > > So it means that during the mapathon, you can see who just makes > occasional slips vs. systematic mistakes. > > At the Cambridge mapathon we used this quite successfully, seeing which > users were putting in a lot of buildings that were not squared (~ 100s of > buildings in two cases). We managed to find them (which of course is not > necessarily trivial), and they corrected the issue before the mapathon was > over. This fixes both the buildings they added (one-off as it were), but > also means they will square buildings in future. > > Another use of the tool is to identify where a particular mapathon as a > whole generates lots of non-square buildings, in which case the supporters > need a bit of additional support :) Also, you can tell who doesn't map much > during the mapathon: They might need a bit more encouragement, to make sure > they don't get frustrated, and to make sure they come back. > > Even though I say this myself, I think the tool has great promise, and we > demo'd this at the last London mid-month mapathon, so some people will have > seen it. We're also doing a webinar on this later this month. The tool is > really just mashup of the RedCross hashtag API and overpass, but I haven't > seen any other tools working like this. (E.g. in osmcha, you can see issues > with changesets, which is great, but afaik it doesn't allow me to see > whether users are make occasional slips, vs. systematic errors.) > > Unfortunately the RedCross api server does not support (see https: > https://github.com/AmericanRedCross/osm-stats/issues/47). Thus we cannot > have an online version, but you need to pull the repo and run locally. > Hopefully this will come very soon. > > The 2nd issue is that it can be hard to find users. I wished that the > tasks manager had a way to register users who are doing a mapathon, i.e. as > you start, you put a table number in, which means helpers can find you on > the basis of your OSM id. > > For your use case, it would be possible to get the script to produce a > list of non-square buildings, which could then load loaded into JOSM, and > cycled through with the TODO plugin... > > Also, any support in our dev efforts is really welcome, including ideas > whether the tool could be integrated into an existing tool... > > Bjoern > > On 15 July 2017 at 13:10, john whelan wrote: > >> I think they need visually inspecting unfortunately. They can be done in >> batches ie look for untagged, then select a
Re: [HOT] Buildings in Chad
I suspect your standards are much higher than mine. Basically I ignore if a building is square or not, in general I ignore buildings unless they have been mapped two or three times in which case if they catch my eye I will delete some of the multiple copies. It might be worthwhile building a script to pick these out I see quite a few of them but you'd need to set up a team to clean them up. If a settlement has been tagged a building then if I notice it I'll retag it landuse=residential. A script might be useful for picking these out, pick out those buildings that are above a predefined size. Again you need someone to inspect them. On buildings I look purely for untagged ways and whilst the todo plugin can be useful sometimes just looking for the untagged by mapper will give you a cluster of untagged buildings that you can tag all at once. The todo plugin is used on the last pass. Mixed in with the untagged buildings will be highways and other features. I have noticed that individual mappers can be very productive. Especially after feedback, whilst maperthons are fine they take up travel time etc. so old aged pensioners or retirees often are less well represented and once they get the hang of it they have the time and patience to map. The other side of maperthons is it takes a fair bit of organising to do one properly and anyone can hold one. From the names of the mappers and the time of day when they map it is doubtful if many mappers first language is English. What I'm looking at is the map after everyone has mapped and the mappers do have different skill levels and training. So for a well organised mapathon your tools will be useful. For the rest of the world I suspect it will be up to other mappers to clean up slowly. Cheerio John On 15 July 2017 at 08:51, Bjoern Hassler wrote: > Hi John, > > At the recent Cambridge mapathon, we used this tool > https://github.com/OSM-Utilities/JOSM-Scripts-HOT/tree/master/www > (which we developed especially for mapathons). > > You enter the hastags and time of your mapathon, and it produces a list of > changes made by those users during the mapathon, that is "validated" > against certain criteria, e.g. whether 4-node buildings are square, whether > there are untagged isolated nodes, etc. The tool gives you a percentage: > > - User ABC: Non-square buildings 2% > ... > - User DEF: Non-square buildings 95% > ... > > So it means that during the mapathon, you can see who just makes > occasional slips vs. systematic mistakes. > > At the Cambridge mapathon we used this quite successfully, seeing which > users were putting in a lot of buildings that were not squared (~ 100s of > buildings in two cases). We managed to find them (which of course is not > necessarily trivial), and they corrected the issue before the mapathon was > over. This fixes both the buildings they added (one-off as it were), but > also means they will square buildings in future. > > Another use of the tool is to identify where a particular mapathon as a > whole generates lots of non-square buildings, in which case the supporters > need a bit of additional support :) Also, you can tell who doesn't map much > during the mapathon: They might need a bit more encouragement, to make sure > they don't get frustrated, and to make sure they come back. > > Even though I say this myself, I think the tool has great promise, and we > demo'd this at the last London mid-month mapathon, so some people will have > seen it. We're also doing a webinar on this later this month. The tool is > really just mashup of the RedCross hashtag API and overpass, but I haven't > seen any other tools working like this. (E.g. in osmcha, you can see issues > with changesets, which is great, but afaik it doesn't allow me to see > whether users are make occasional slips, vs. systematic errors.) > > Unfortunately the RedCross api server does not support (see https: > https://github.com/AmericanRedCross/osm-stats/issues/47). Thus we cannot > have an online version, but you need to pull the repo and run locally. > Hopefully this will come very soon. > > The 2nd issue is that it can be hard to find users. I wished that the > tasks manager had a way to register users who are doing a mapathon, i.e. as > you start, you put a table number in, which means helpers can find you on > the basis of your OSM id. > > For your use case, it would be possible to get the script to produce a > list of non-square buildings, which could then load loaded into JOSM, and > cycled through with the TODO plugin... > > Also, any support in our dev efforts is really welcome, including ideas > whether the tool could be integrated into an existing tool... > > Bjoern > > On 15 July 2017 at 13:10, john whelan wrote: > >> I think they need visually inspecting unfortunately. They can be done in >> batches ie look for untagged, then select a mapper, then zoom in to a >> settlement so sometimes you can clear thirty at once but I can't think how >> you'd scrip
Re: [HOT] Buildings in Chad
Hi John, At the recent Cambridge mapathon, we used this tool https://github.com/OSM-Utilities/JOSM-Scripts-HOT/tree/master/www (which we developed especially for mapathons). You enter the hastags and time of your mapathon, and it produces a list of changes made by those users during the mapathon, that is "validated" against certain criteria, e.g. whether 4-node buildings are square, whether there are untagged isolated nodes, etc. The tool gives you a percentage: - User ABC: Non-square buildings 2% ... - User DEF: Non-square buildings 95% ... So it means that during the mapathon, you can see who just makes occasional slips vs. systematic mistakes. At the Cambridge mapathon we used this quite successfully, seeing which users were putting in a lot of buildings that were not squared (~ 100s of buildings in two cases). We managed to find them (which of course is not necessarily trivial), and they corrected the issue before the mapathon was over. This fixes both the buildings they added (one-off as it were), but also means they will square buildings in future. Another use of the tool is to identify where a particular mapathon as a whole generates lots of non-square buildings, in which case the supporters need a bit of additional support :) Also, you can tell who doesn't map much during the mapathon: They might need a bit more encouragement, to make sure they don't get frustrated, and to make sure they come back. Even though I say this myself, I think the tool has great promise, and we demo'd this at the last London mid-month mapathon, so some people will have seen it. We're also doing a webinar on this later this month. The tool is really just mashup of the RedCross hashtag API and overpass, but I haven't seen any other tools working like this. (E.g. in osmcha, you can see issues with changesets, which is great, but afaik it doesn't allow me to see whether users are make occasional slips, vs. systematic errors.) Unfortunately the RedCross api server does not support (see https: https://github.com/AmericanRedCross/osm-stats/issues/47). Thus we cannot have an online version, but you need to pull the repo and run locally. Hopefully this will come very soon. The 2nd issue is that it can be hard to find users. I wished that the tasks manager had a way to register users who are doing a mapathon, i.e. as you start, you put a table number in, which means helpers can find you on the basis of your OSM id. For your use case, it would be possible to get the script to produce a list of non-square buildings, which could then load loaded into JOSM, and cycled through with the TODO plugin... Also, any support in our dev efforts is really welcome, including ideas whether the tool could be integrated into an existing tool... Bjoern On 15 July 2017 at 13:10, john whelan wrote: > I think they need visually inspecting unfortunately. They can be done in > batches ie look for untagged, then select a mapper, then zoom in to a > settlement so sometimes you can clear thirty at once but I can't think how > you'd script that. > > It's just a little frustrating that some mappers have mapped so much > without being given feedback that somehow they weren't adding tags and > feedback more than a week or two after the event gives diminishing returns. > > I think the big message is if your project is in Chad pull the map in > again in a couple of weeks time and it should be a lot better. > > Cheerio John > > On 15 July 2017 at 03:19, Bjoern Hassler wrote: > >> Hi John, >> >> Let me know if I can do anything scripting-wise. >> >> Bjoern >> >> >> On 15 July 2017 at 02:49, john whelan wrote: >> >>> I'm looking at Chad by bringing 10% of the country at a time. I'm >>> seeing thousands of untagged ways most of which are buildings, far more >>> than I would expect. >>> >>> If you're running a project in Chad that includes mapping buildings you >>> may wish to check the map in a week or two as I continue to tag untagged >>> ways including buildings. >>> >>> You may also want to see if you can have someone validate your Chad >>> projects with JOSM since its much easier to avoid problems by giving >>> feedback than it is to clean up afterwards. >>> >>> Cheerio John >>> >>> ___ >>> HOT mailing list >>> HOT@openstreetmap.org >>> https://lists.openstreetmap.org/listinfo/hot >>> >>> >> > ___ HOT mailing list HOT@openstreetmap.org https://lists.openstreetmap.org/listinfo/hot
Re: [HOT] Buildings in Chad
I think they need visually inspecting unfortunately. They can be done in batches ie look for untagged, then select a mapper, then zoom in to a settlement so sometimes you can clear thirty at once but I can't think how you'd script that. It's just a little frustrating that some mappers have mapped so much without being given feedback that somehow they weren't adding tags and feedback more than a week or two after the event gives diminishing returns. I think the big message is if your project is in Chad pull the map in again in a couple of weeks time and it should be a lot better. Cheerio John On 15 July 2017 at 03:19, Bjoern Hassler wrote: > Hi John, > > Let me know if I can do anything scripting-wise. > > Bjoern > > > On 15 July 2017 at 02:49, john whelan wrote: > >> I'm looking at Chad by bringing 10% of the country at a time. I'm seeing >> thousands of untagged ways most of which are buildings, far more than I >> would expect. >> >> If you're running a project in Chad that includes mapping buildings you >> may wish to check the map in a week or two as I continue to tag untagged >> ways including buildings. >> >> You may also want to see if you can have someone validate your Chad >> projects with JOSM since its much easier to avoid problems by giving >> feedback than it is to clean up afterwards. >> >> Cheerio John >> >> ___ >> HOT mailing list >> HOT@openstreetmap.org >> https://lists.openstreetmap.org/listinfo/hot >> >> > ___ HOT mailing list HOT@openstreetmap.org https://lists.openstreetmap.org/listinfo/hot
Re: [HOT] Buildings in Chad
Hi John, Let me know if I can do anything scripting-wise. Bjoern On 15 July 2017 at 02:49, john whelan wrote: > I'm looking at Chad by bringing 10% of the country at a time. I'm seeing > thousands of untagged ways most of which are buildings, far more than I > would expect. > > If you're running a project in Chad that includes mapping buildings you > may wish to check the map in a week or two as I continue to tag untagged > ways including buildings. > > You may also want to see if you can have someone validate your Chad > projects with JOSM since its much easier to avoid problems by giving > feedback than it is to clean up afterwards. > > Cheerio John > > ___ > HOT mailing list > HOT@openstreetmap.org > https://lists.openstreetmap.org/listinfo/hot > > ___ HOT mailing list HOT@openstreetmap.org https://lists.openstreetmap.org/listinfo/hot
