Ben
I also forgot that for this specific case in CAR, we will use Bing and the Task
Manager. I agree that this is a good approach to validate WRI imports against
Bing. This is the best way to do it and try to reconcile the OSM and the WRI
databases. This will also help to evaluate the quality of the WRI data.
Anyway, It is important to insist that we should not favor import data blankly
just because their is nothing in the area.
Pierre
>________________________________
> De : Ben Abelshausen <[email protected]>
>À : Pierre Béland <[email protected]>; "[email protected]"
><[email protected]>
>Cc : Severin MENARD <[email protected]>; Andrew Buck
><[email protected]>
>Envoyé le : Mardi 2 juillet 2013 15h04
>Objet : RE: [HOT] Central African Republic import of WRI data
>
>
>
>Hi,
>
>Many of these things have been discussed and considered. This is why we
>suggested this new way of using the tasking manager, the wri data and bing
>together and extract the useful data. The step of manually reviewing the wri
>data against the imagery and using it as a guide instead of a blind import is
>exactly why this task was created...
>
>It is difficult to know how to combine these different views/rules and
>guidelines and to apply them to this situation. I hope everyone agrees there
>is some potential to work this way, we just need to figure out how :)
>
>Regards,
>
>Ben
>
>________________________________
> From: Pierre Béland
>Sent: 2/07/2013 19:45
>To: [email protected]
>Cc: Severin MENARD; Andrew Buck; Ben Abelshausen
>Subject: Re: [HOT] Central African Republic import of WRI data
>
>
>To my point of view, you should never import blankly. If its just a problem of
>accuracy, I agree that with a 200 meters error in a desertic area, it may
>still be helpful to import. But other then that, if we are not sure about the
>quality of the data, if the roads exist really, we should not import just to
>replace blank spots.
>
>
>I have too often seen roads imported for Africa that correspond to nothing on
>the ground. I have tried to follow these roads in mountains and passed long
>time to evaluate thee. And since we dont have that many mappers, and that it's
>time consuming to evaluate the accuracy of such long roads crossing countries
>and correct / delete these, we often end up with these incorrect roads sitting
>in the database because the contributors do not want to pass that many time
>evaluating and correcting this.
>
>Therefore, it is important to take time to plan such imports, to take time to
>evaluate accuracy. It is sometime better to have nothing then anything. This
>indicates that some work is still to be done. We dont map just to cover the
>map. It take much less time to add new roads with good quality data then to
>try to evaluate and correct bad data imports.
>
>
>
>
>Pierre
>
>
>
>>________________________________
>> De : Ben Abelshausen <[email protected]>
>>À : Andrew Buck <[email protected]>
>>Cc : Severin MENARD <[email protected]>; "[email protected]"
>><[email protected]>
>>Envoyé le : Mardi 2 juillet 2013 12h21
>>Objet : Re: [HOT] Central African Republic import of WRI data
>>
>>
>>
>>Andrew,
>>
>>Actually I agree, but I discussed this before with some other OSM/HOT people
>>(It's been so long I don't remember when or who) and it was said that using
>>inaccurate data might be worse than nothing.
>>
>>
>>For the CAR specifically I think using the data might be better than not
>>using it. There is nothing there, no data at all, for most of the country.
>>There won't be a mapping community there in the near future... etc.
>>
>>
>>
>>Met vriendelijke groeten,
>>Best regards,
>>
>>Ben Abelshausen
>>[email protected]
>>http://twitter.com/xivk
>>
>>
>>On Tue, Jul 2, 2013 at 6:09 PM, Andrew Buck <[email protected]> wrote:
>>
>>Regarding areas with no imagery. I would still say to import the roads. A
>>couple hundred meters off is a lot, but it is better to have the
>>'connectivity' mapped (i.e. what town connects to what other town) than to
>>have nothing at all. Someone following the road in a car does not mind that
>>their GPS tells them they are "beside" the road. Also, it is my
>>understanding that the data has some road names and such so I would import
>>them to get that. They can always be refined later when bing becomes
>>available. The imagery is being added fast, and just in the last year we
>>have gone from only a few cities in Africa to much of the continent being
>>covered. I expect much of the rest to come online shortly.
>>>
>>>
>>>Importing some of the data now and the rest later brings its own problems,
>>>so I would suggest putting it in.
>>>
>>>
>>>-AndrewBuck
>>>
>>
>>_______________________________________________
>>HOT mailing list
>>[email protected]
>>http://lists.openstreetmap.org/listinfo/hot
>>
>>
>>
>
>
_______________________________________________
HOT mailing list
[email protected]
http://lists.openstreetmap.org/listinfo/hot
