[Talk-GB] Importing Website Data

Warin 61sundowner at gmail.com
Tue Dec 19 01:16:39 UTC 2017


On 19-Dec-17 05:42 AM, Ilya Zverev wrote:
> 18.12.2017 19:14, Frederik Ramm пишет:
>> Hi,
>>
>> On 12/18/2017 03:09 PM, SK53 wrote:
>>> Personally, I'd also be chary of turning OSM into a repository of
>>> scraped data rather than one of surveyed geodata.
>>
>> And more: the more imports you do, the more OpenStreetMap becomes an IT
>> project where computer nerds script, collect, convert, conflate, and
>> interpolate away, instead of a project where all sorts of people from
>> all walks of life contribute their knowledge.
>>
>> I think we're not an IT project, and that's good.
>
> We are partly an IT project, which you cannot take away. Also, the 
> number and size of imports does not affect how OSM is perceived.
>
> For example, OSMers map very few POIs (relatively), and after they do, 
> they update virtually zero of them. Not counting a few active mappers 
> in UK and Germany.
>
> To think that you can import everything is to overstate the quantity 
> of open data in the world. Open data is mostly points of interest. In 
> some countries it's also roads. And that's all. OSM is so much bigger 
> than that.
>
> As for importing POIs, I think that needs to be done, because no 
> mapper group would be able to map all fuel station, all Tesco shops, 
> or all museums. And if they do, in ten years, but that time half of 
> what they've collected will be obsolete. The choice is not between 
> manual mapping and importing, it's between importing and not having 
> the data ever.
>
> And since there isn't open data on everything, even if you import all 
> POIs you can possibly have in machine-readable format, that will still 
> be at most 10% of all POIs, even in UK or Germany. Plenty to map by 
> going outside.
>
> So even an army of IT folks armed with data scrapers and contracts 
> with every data aggregator would be no match to common mappers, armed 
> with a pen and a camera. I fail to see what you're afraid of.

Two divergent points of view if you go to the extremes. Here are some 
extremes;


A) Only physically survey data is for OSM.
Not possible to physically survey some things like National Park 
boundaries in Australia -  where they are marked the marks are well 
inside the actual boundary. A physical survey will result in large errors.
B) Only open data is to be used.
Some of this is out of data - an on the ground physical survey will 
confirm it.

The best for me is a mix of sources, with a preference for the on the 
ground physical survey. However I can map a lot more area from open 
source data quickly and accurately that I can from a physical survey. In 
some cases of tree cover or GPS reflections, satellite imagery is more 
accurate that GPS surveys. So a mix of sources for me.



More information about the Talk-GB mailing list