Hi,
I've been playing with the data and checking this DAE/EAD data and it's not so easy after all. There are several shortcomings in this dataset, I'll list them up. But in general, it's not wel structured and it's mixed languages as well. See url [3] for the files. I was going to make an attempt, but it's a lot of work in the first place on this source data which really sucks quality wise: - Names are often in capitals. (and so are plenty of other fields) - Counting the number of distinct provinces, we have more than fifty - It's not limited to Belgium, I had to clean out FR/NL/AF(?) countries - There is verbose information, like floors and location descriptions who are a bit hard to structure, the language is mixed, so I tried searching for different typical words in an attempt to split it up into :fr and :nl location descriptions - addressing scheme is not consistent, sometimes numbers before streets. It's also in 1 value field, so splitting up the numbers from the streetinfo willl take some more serious regex-work/edit. - Some fields are cutoff : <tag k='Batiment' v='Zaal De Bo' /> <tag k='Nom' v='De Bond Val-Meer' /> - 4 identical duplicate entries - 125 non-identical duplicate entries , different housenumbers/wrong streetnames too. I had to juggle values between key's as well to make it a bit more consistent, but it's just a best effort. It's full of weird values like : level=ziekenwagen of level=garage (probably mean -1), level=Paal.... Also, Rez de chaussée in all it's variants: rez, Rdc, Rez, Etc... and in dutch: glvl, GV. So this list contains mobile assets as well! And still there is interesting stuff in there like the location (indoor/outdoor), access (private/public) and what's also nice, opening hours if available. But last, and not least: - The GPS coordinate accuracy is really terrible, I did some random checks of places I know and some of the locations are in the middle of crossroads or contain an address + number that is way vs. using OSM addresses (vs. raw CRAB too). In fact, I think I only found 1 that I believe is about correct (Nekkerhal Mechelen) , only 1 of 3 in that area in fact. So, We can't really trust this data... We can't import this. At best we could use it to compair OSM data against, just to get an idea. Or you could use it to compare with what you find in the field. My work files can be downloaded using the links below [3]. You can open it up in JOSM. So You can check for yourself. Please don't force-upload this. To get the existing defibrillators use this overpass query [1] , there are ways and nodes present containing the tag. But the emergency key on a way's is an error. [2] Feel free to check out how far I got in trying to get some structure here [3] Glenn [1] http://overpass-turbo.eu/s/gDA [2] http://wiki.openstreetmap.org/wiki/Tag:emergency%3Ddefibrillator [3] http://aed.byteless.net/ _______________________________________________ Talk-be mailing list [email protected] https://lists.openstreetmap.org/listinfo/talk-be
