There are many cool points listed at Atlas Obscura. I think it might be useful to have some link between corresponding OSM place and Atlas Obscura (if any). Of course, as seen from the terms Terms of Use - Atlas Obscura, I cannot reuse the data from Atlas Obscura in OSM, but it should be fine to have something like atlas_obscura=yes or maybe better website:atlas_obscure=<url>.
Do you think it makes sense, how can it be done?
And maybe more general question as well - there are many interesting websites about the places which are not “official websites”, do you think it might make sense to add similar links for the ones which are useful from mapper point of view?
There are a number of keys that serve as deep links into a specific source, but these sources are typically databases that that serve as authority files for places, so they fall under the ref:*=* or heritage:website:*=* namespace. Atlas Obscura is more of a travel guide than an authority file. The closest documented key I know of is kudish=*, which references a chapter of a history of railways in a region of New York. Taginfo also shows website:trailforks=*, which is specific to the Trailforks website.
Incidentally, every place described in Atlas Obscura has a Wikidata item tagged with an Atlas Obscura ID (P7772) property. Here are some queries for OSM elements that represent places featured in Atlas Obscura via the three main SPARQL services:
This demonstrates the usefulness of wikidata=* as a general hub for external references. You can even use Hub to hop directly from Atlas Obscura to OSM and vice versa.
There are an almost infinite number of interesting websites, and linking to all of them seems to make no sense to me. Links to the website of the place itself and Wikipedia are a good idea, but we should stop there.
This actually sounds really good, but I would still imagine it would be the best to be able to query from OSM directly, e.g. via overpass.
Why won’t we have atlas_obscura_id=P7772 as a tag in OSM directly? Especially when, as shown, this is notable enough (and not “some random website”)
Well, linking to the ones which mapper considers useful for them personally and for at least some part of the users. It’s absolutely the same as having some very specific tags in OSM already which I personally think are absolutely useless (at least, to me), but I’m tolerant here and there might be people interested in different types of data.
Anyone is free to add anything he thinks is useful, but I’m wondering where this is going to end. I had never heard of Atlas Obscura until your post, so I can’t judge how notable it is. I have heard of other travel sites like Tripadvisor, Wikivoyage, Wikitravel, Lonely Planet, etc. Should links to these sites then also be added to the OSM place? And maybe add links to traildino, wikiloc and alltrails for hiking trails on OSM? And a link to someone’s travel blog who visited the site? …
Actually I’d say “yeah sure, why not?”. I don’t see any problems having potentially useful information about places, OSM is a database, right? On Wikidata there are a lot of IDs from different websites, some of them are already present or might be present in OSM (like Twitter/X, Facebook), so why not add another one?
As for “random blogs” it’s a bit different, I don’t see anything which might be a list (because dedicating a special key for each “random blog” is for sure too much). But if someone would to suggest some list-like value personal_blogs_urls=<article1>;<article2>;...;<articleN> - I do see some value here (though not as much as more notable websites mentioned above).
Bigger problem with URLs in general that they tend to die quite soon (one can see this a lot in “image” URLs where tons of links are dead), but this is something completely different and off-topic here…
How do you envision someone using a query for Atlas Obscura–tagged elements? All three of the SPARQL services I mentioned can plot the featured places on an OSM map, just like Overpass turbo, and have APIs that can return these places within seconds. In my opinion, the fact that the queries rely on Wikidata would be an implementation detail from an end user perspective. This is not to say that we can’t directly tag url=* or ref:atlas_obscura=* in OSM, but if we reinvent this wheel, someone would have to maintain it.
I clicked thru all three links you provided hoping I can quickly export those to .GPX to import to my phone (e.g. Locus). I didn’t find the way to get GPX from the services you provided, as opposed to Overpass where it’s available in one click…
As preparation for the trip - export to GPX and use it in your personal map during the trip - that’s at least would be my main use case.
As part of the trip, having their map around and asking the question "Hey, what are the places which are in Atlas Obscura around me` (that’s what official app already is capable of, but it’s online and, well, separated app. If I could query it directly in offline dump e.g. in my Locus, I don’t even need any extra apps. Links with wikidata here are not really useful, if we’re talking about offline case.
As part of figuring out details about the place one has found randomly “wow, that was a huge Peps bottle at this gas station, can I find more details about it”
As some extra information already provided by Organic Maps and Osmand, someone has mentioned they try to show some available Wikipedia data already today (“live map” in Osmand? Not sure, haven’t used it at all). While the content of the non-Wikipedia websites cannot be shown directly as copyrighted, the apps can always show some extra links “hey, this place also has the page on Atlas Obscura, here is the link, go check it out”
In QLever, click on “Map view” to open the results in Petrimap, then click on the “Export as GeoJSON” button to export a GeoJSON file. Overpass turbo’s GPX export option uses the togpx library to convert the data from GeoJSON to GPX. This library is also available as a command line tool, or you can use any online converter or desktop GIS tool such as QGIS.
This may not be as convenient as the one-click GPX option in Overpass turbo, but someone could trivially build a webpage that ties it all together.
Yes, if you select a location on the map, OsmAnd links to Wikipedia and embeds Wikimedia Commons images, based on Wikidata if the image isn’t explicitly tagged on the element. Organic Maps also links to some travel websites, though not without controversy.
Adding website:atlas_obscura= tags in OSM makes sense for useful external links. Similarly, adding links to other informative sites can enrich OSM data for mappers and users alike.
I quite like Atlas Obscura and I am on board with the goal (seeing Atlas Obscura places in OSM-based maps and apps), so I looked a bit more into the the solution proposed by @Minh_Nguyen of going via Wikidata.
Unfortunately it’s not the case that every Atlas Obscura item is linked to from Wikidata. There are over 27,500 places on Atlas Obscura (according to Atlas Obscura). Only 11,872 items on Wikidata have the atlas obscura property and Minh’s queries return about 10,200 OSM objects that link to Wikidata items that in turn link to Atlas Obscura places.
So only about a third of Atlas Obscura places can be found indirectly from OSM by going via Wikidata, and that’s not because the OSM objects are missing Wikidata tags (which we would have tools for), it’s because the Wikidata objects don’t link to Atlas Obscura places, or because the Wikidata objects don’t even exist yet for many Atlas Obscura places…
That seems solvable. These missing places would satisfy Wikidata’s criteria for inclusion by virtue of having an Atlas Obscura article (and typically for other reasons besides). Wikidata welcomes bulk edits. Tools like OpenRefine and QuickStatements can help us make quick work of it.