[HOT] paths, tracks and unclassified in West Africa

Thomas Gertin tgertin at gmail.com
Thu Jul 16 14:29:07 UTC 2015


Some great input is being provided. Primary / secondary / tertiary highways are not encountered very often in a typical HOT project by the average mapper. Therefore even though they are very important to classify and may be mentioned in the training material, they should not be emphasized. Classifying these highway types is better suited for a ‘meta’ task where a mapper would look at a larger zoomed out area and gain better context for classifying these highway types by analyzing the sizes of the urban areas that are connected by them. In addition, on the ground validation would always be great to have as well. 

I agree that more validation is greatly needed on HOT projects in general. I think the greatest challenge is how to accomplish this objective; there are simply not enough people doing it. Maybe we can incentivize validation with badges.

I think if HOT wants to endorse regional tagging schemes for HOT projects, then West Africa or Africa in general would be a good place to start.

Thanks,

Tom G


> On Jul 16, 2015, at 9:32 AM, john whelan <jwhelan0112 at gmail.com> wrote:
> 
> Validation can mean many things, in a HOT context to me it means going in and correcting the glaring errors.  Highway classification or missclassification is subjective as has been stated so I would not include putting the correct tags on Primary / secondary / tertiary as part of the primary role of the validator.  I would say changing highway=pedestrian to highway=path in a rural area of West Africa would be part of validation.
> 
> I work with three others who do validation, there aren't many of us who do it if you look through the projects there are very few that are validated completely and setting the bar to high means fewer people will do it.  Personally I think we need more validation whether that should be two passes, one a less experienced validator and one a more careful validation is one open to debate.  On one project I did a quick and dirty validation that picked up 80% of the errors and it was suggested that I should have done a more complete validation.  It's a judgement call, my feeling was the quality and reliability of the mapping was better after a quick and dirty validation which was not to my normal validation standard than without it.
> 
> In my opinion OSM mapping with no resource constraints often is done to a high quality standard, in HOT mapping we don't have enough trained and experienced mappers and validators to map to the standard we would like to have the maps mapped to in the time that the clients would like the map. 
> 
> So what can we simplify?
> 
> Cheerio John
> 
> On 16 July 2015 at 09:13, Robert Banick <rbanick at gmail.com <mailto:rbanick at gmail.com>> wrote:
> That makes sense. Would you suggest putting road classification into the validation stage then? Or have a classification stage in between?
> 
> On Thu, Jul 16, 2015 at 9:03 AM, john whelan <jwhelan0112 at gmail.com <mailto:jwhelan0112 at gmail.com>> wrote:
> The suggestion is not that Primary / secondary / tertiary should not be mapped, often when the HOT mappers start the major highways are tagged Primary / secondary / tertiary the suggestion is to simplify guidance to new or inexperienced mappers.  76% of HOT Nepal mappers mapped for an hour or two and that was it.
> 
> I don't think we can afford to give them four hours training in how to classify a road, there would be no time left for mapping.
> 
> For these sort of highways then map something and let someone else upgrade the tag to Primary / secondary / tertiary is my suggestion.
> 
> Cheerio John
> 
> On 16 July 2015 at 08:55, Robert Banick <rbanick at gmail.com <mailto:rbanick at gmail.com>> wrote:
> Hey all,
> 
> Speaking as a humanitarian GISer who's used HOT road layers quite a bit in a few crises, the road classifications really help. Primary / secondary / tertiary are useful, albeit vey incomplete, measures of the importance of roads that we can use to eyeball transit times etc. I would be strongly against ignoring those classification tags. I do agree we need more consistency in how they're applied however.
> 
> Perhaps we can have general regional guidelines and then someone gets charged with developing a country-specific taxonomy for any major activations?
> 
> Best,
> Robert
> 
> On Thu, Jul 16, 2015 at 8:44 AM, john whelan <jwhelan0112 at gmail.com <mailto:jwhelan0112 at gmail.com>> wrote:
> Given that HOT mappers often do not have a PhD in African road classification and it appears to be subjective perhaps we can come up with a useful simplified interpretation or guidelines for inexperienced mappers?
> 
> My thoughts would be to suggest that mappers in general ignore primary, secondary, tertiary, classifications, if the road is mapped then a local or classification specialist can tag with one of these if required.
> 
> Cheerio John
> 
> On 16 July 2015 at 00:23, Thomas Gertin <tgertin at gmail.com <mailto:tgertin at gmail.com>> wrote:
> I am adding to the discussion of highway tagging in West Africa. All of the projects that mapped highways in West Africa that I have seen or been a part of followed the guidance of the Highway Tag Africa wiki page (http://wiki.openstreetmap.org/wiki/Highway_Tag_Africa <http://wiki.openstreetmap.org/wiki/Highway_Tag_Africa>).
> 
> This past Spring I worked with some colleagues to create this tracing guide (http://hotosm.github.io/tracing-guides/guide/liberia.html <http://hotosm.github.io/tracing-guides/guide/liberia.html>) for mapping River Cress and Grand Gedeh Counties in West Africa. The tracing guide was based on our interpretation of the Highway Tag Africa wiki page. This tracing guide is quite good, and mappers appreciated the pictures and GIFs that show examples.
> 
> When building the tracing guide I came to a few conclusions. When reading the Highway Tag Africa wiki page I felt it have been wrong for me to alter the instructions. It would have resulted in inconsistent tagging in the region. I trust that a good amount of research and discussion has taken place to get it to the point it is now.
> 
> - The guidance in the wiki could have been clearer. Although I notice that is has improved since even a few months ago, there are now some example pictures in there.
> 
> - It is difficult to teach someone how to classify highways. There are eight types and often it is not clear when deciding between primary, secondary, tertiary, and unclassified highways because the only difference between them is the subjective size of the urban areas that are connected by them.
> 
> - The unclassified road type was unintuitive the first time I read the Highway Tag Africa wiki page. To me unclassified means something that has no classification. Yet in the Highway Tag Africa wiki page it clearly has a classification. I think the term ‘unclassified’ means something else in other places though.
> 
> I think having pre-set tags available as a plugins to iD editor should be a HOT goal, if it isn’t already. I don’t think we need there to be a universal tagging set. People who set-up projects on the Tasking manager could define the tags that fit best for the project. Although I think it would be useful to further standardize some tags across many geographical areas; it is important to maintain the flexibility for the geographical areas that need unique tags.
> 
> Thanks,
> 
> Tom G
> 
> _______________________________________________
> HOT mailing list
> HOT at openstreetmap.org <mailto:HOT at openstreetmap.org>
> https://lists.openstreetmap.org/listinfo/hot <https://lists.openstreetmap.org/listinfo/hot>
> 
> 
> 
> _______________________________________________
> HOT mailing list
> HOT at openstreetmap.org <mailto:HOT at openstreetmap.org>
> https://lists.openstreetmap.org/listinfo/hot <https://lists.openstreetmap.org/listinfo/hot>
> 
> 
> 
> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20150716/b62e2333/attachment.html>


More information about the HOT mailing list