[OHM] Wikimaps reporting

[Robert Warren]

Tim,

Geo referencing should work without problems with maps that are… actually surveyed. We may end up with several species of sea monsters on the OHM before we are done. I've used qgis to do the re-projection, through beware that some versions are buggy.

I've had some successes in vectorizing trench maps - in the end the lesson is that generating vectors from map imagery is fairly simple; it's the recognition of the map feature type that is difficult,  especially when the features overlap. There is a crowd sourced way around it that increases throughput by about 50x over manual methods but I have not had time to harden the code for public use.

Something that I would like the Wikipedia Commons to generate is a linked open data interface to its data, a la dbpedia. Parsing copyright and description blocks is brittle and a 'machine readable' interface would allow us to process more data using automated means and cross link it.

While I am playing the pot calling the kettle black, OHM will be getting its own linked geo data feed soon. Still fighting with tile renderer for temporal support and rendering trenches.

All my best,
Rob


On 2013-05-31, at 8:00 AM, historic-request at openstreetmap.org wrote:

> From: Tim Alder <tim.alder at s2002.tu-chemnitz.de>
> Subject: Re: [OHM] Wikimaps reporting
> Hello,
> I hope it's a kind of misunderstanding. There is nothing inside Wikimaps 
> that could be a show-stopper for OHM.
> 
>  We want to scan and publish historical maps at Commons. If we are 
> successful with the georeferencing process, theses maps could be as a 
> side-effect included in an OSM-editor as base for OHM. If not, an other 
> project can fill this piece of the project. (But remember with Commons 
> we have an media archive and a community to handle over 17 million 
> images that's something what OSM not have. What we also have is a 
> relative good contact to libraries and museums.)
> 
> In the second step we will perhaps discuss like everybody else about the 
> best way to vectorize stuff, but we will not dictate anything. If we 
> have not better ideas, perhaps we also only following the process of OHM.
> 
> In the third step we want to use OHM-data like we do this with OSM-data 
> since 2009 at toolserver and we want to bring it to Wikipedia in similar 
> way like the WIWOSM project or in other ways. I can't see how this 
> should be a problem for OHM.
> 
> But what I see is that we can bring lot's of historian to the project 
> because we have inside Wikipedia thousands historical article?s that 
> waiting for better maps.
> 
> I hope this is an acceptable way for a cooperation.
> 
> Greetings Tim alias Kolossos
>