Latest comment: 1 year ago2 comments2 people in discussion
@Prototyperspective, I haven't used the apps yet but I like the concept of this wish. I don't expect you to reveal your location or go exploring, but are there any screenshots out there of how crowded the current maps get in popular areas?
Commons app: Nearby
You mentioned WikiVoyage so I will mention Commons. From what I can tell, the Commons app has a nearby activity feature that can filter between places that exist and places needing a photo. Filtering alternatively with the suggestions in the wish would be great. Eg "show me nearby places with important articles that are inadequately illustrated". Commander Keane (talk) 02:38, 13 November 2024 (UTC)Reply
Recommending using the app. Good question! I couldn't find one now except for the video on the right but will check again later – it's difficult to demonstrate this with just a screenshot however, it needs to be either a video or a modified image where each dot is categorized by type. I found many areas are cluttered with items about train stations, schools, graveyards, and so on. I get that some people are interested in these things or that it's useful to some but a filter is needed so people can enable/disable types of articles depending on what they use this for (and I think most use this for exploring new places)…and both for areas that have lots of items as well as areas that have few items so one quickly see interesting places in a large region without it being buried underneath 10 or so rather trivial places. A demo image would probably be useful so maybe I'll create on with using a screenshot at some point.
By the way, a further way to improve the usefulness would be to set one primary language and one or several ancillary ones so it shows all items in your native language except those that are only available in some other language like the language of the foreign country one is visiting so it also shows those places on the map (especially useful if combined with machine translation as proposed in other wish(es)).
It doesn't have to be crowded for this to be useful/needed – I think it would make a lot of sense to enable users to filter it so that they can identify places to go to at the scale of a large city. This would work best if there's also filters for things that consider WikiProject importance ratings and/or article view count (and/or similar things) which are generally higher for quite notable things like museums, ancient buildings, bridges with good views, recreation facilities, and sightseeing spots than let's say one for a subway station.
Good point – yes such filters would also be useful for the Commons app map. There are some really major issues to fix there before that I think, like enabling videos and audio files to be played in the app. I don't know if the Commons app uses or could use the Places map – this may make it easier to implement such things / adopt things that are implemented for the WP app Places map. When it comes to photos of places, I think the main type of media missing on Commons are illustrations and similar images, not photos of places so I think while useful this wouldn't be important to improve upon.
Another note: I think the clustering feature makes it also really cumbersome to use the Places map and often not usable in the real world – see phab:T360213. Adding for example an option to change this (which is possible per MapLibre devs) would be synergetic with what's proposed here.
Lastly, one could eventually extract an app with just the Places map as well as the WikiVoyage pages as a WikiVoyages app and it could be quite successful and make the project be used much more. Prototyperspective (talk) 12:04, 13 November 2024 (UTC)Reply
This is also related to a difficulty: if the filters are implemented via deepcategory then these may not work because deepcategory breaks at a specific size of categories. Additionally, there often is some miscategorization due to which unrelated articles are underneath some broad category. However,
even when there's lots of offtopic articles, there may be more or less no offtopic ones that have geocoordinates.
Nevertheless, using deepcategory (or even any dynamic use of Wikipedia categories more broadly) may not be the best approach or at least would be insufficient as it may also need some caching or similar. → Another approach would be to create datasets of articles for the filters where there could be more and more available filters over time. These could be built in various ways such as with SPARQL and the Quarry query service and even cleaned to remove offtopic articles. So instead of dynamic loading using deepcategory or some query, it would load from this dataset of the filter. This is just the rough outline of the idea – see https://zenodo.org/records/18955341 "ArtefactsLocation Database" used in the aforementioned app. I don't know how it was built; maybe the user could elaborate or it's explained on that Zenodo/GitHub site. Prototyperspective (talk) 22:38, 11 March 2026 (UTC)Reply
Thanks for the mention.
I should definitely update the dataset with a description of how this database was built. I'll do it briefly here.
Take, for example, the wiki category "Gothic architecture of France".
This category has sub-categories and "end" wiki pages with buildings themself.
The program passed through this category and all subcategories, typing a list of artifacts with their synonyms in a dozen languages, with architects and dates of creation, etc. (if they were indicated in the corresponding block of the wiki page)
All objects obtained in this way fell under the general category of "Gothic Architecture" (there is, of course, such a general category on Wikipedia).
Thus, by combing through countries and styles/categories, the database was built five years ago.
This construction was the essence of it: the goal was to obtain objects clearly assigned to a specific architectural category (or several at once) and with a more or less adequate description.
Otherwise, one could have skipped the hassle and used Wikidata, etc.
Now I'm only busy updating the database using the same method (articles appear) and monitoring its accuracy: pages can disappear, names can change, duplicates can appear, etc.
I typically perform these checks once a year, which can take up to a week/10 days of machine time. The number of "errors" (name corrections, duplicate deletions, etc.) does not exceed 0.3% of the database size—at least, that's been the case for the last three years.
Over the past year, I've switched to "semi-manual work": for the first time, I've added to the database from sources other than Wikipedia (this only applies to Munich) and am expanding it by working my way through the list of architects of a particular style (in Wikipedia, of course). Often, a given architect's list of works has more links to existing buildings in Wikipedia than the corresponding category. But this work is endless and, as I mentioned, essentially manual. The share of such additions to the database does not exceed 0.02% of its volume. Myropolskyi (talk) 13:13, 12 March 2026 (UTC)Reply
Thanks for this info! Does the tool use Wikipedia infoboxes, Wikipedia categories of the 'end' articles and/or Wikidata statements of the article's WD item to get this structured data?
More info how this was built at the dataset page would be very useful, thanks.
If you're saying the dataset was built 5 years ago, does it not have data from for example articles create or data added to existing articles/items in the last 5 years?
Did you have miscategorization issues where say there were articles about films underneath cat Gothic architecture and how did you solve this? Is that what you're referring to with I typically perform these checks once a year?
The program passed through this category and all subcategories, typing a list of artifacts with their synonyms in a dozen languages, with architects and dates of creation, etc. This is the most relevant part to this wish – the same could be used for creating the filters without being limited by deepcategory-search-operator-limitations or miscategorization-issues. Further info regarding it in relation to potential application for Nearby Map Filters would be quite valuable here I think.
Otherwise, one could have skipped the hassle and used Wikidata, etc. So it's because as I've pointed out and discussed here, Wikipedia often has more data (items and structured data) than Wikidata or are you referring to sth else with "Otherwise"?
Often, a given architect's list of works has more links to existing buildings in Wikipedia than the corresponding category That's a general problem on Wikipedia. Wikidata isn't much of a solution as it often even misses even items that have an article and are in the Wikipedia category. What could help is tools that check lists to then automatically suggest a category for the articles in the list where the cat is missing. Prototyperspective (talk) 13:32, 12 March 2026 (UTC)Reply
> Does the tool use Wikipedia infoboxes,
exactly.
Then you need to normalize the dates (semi-manually). For example, a date from infobox like "created 1154 rebuilt 1678" becomes "creation completed" 1154-1678.
> does it not have data from for example articles create or data added to existing articles/items in the last 5 years?
It is more or less fresh. There are fields in each table: created, modified, reviewed, so I have control. Last "stroll" through the whole wiki was 1/2 year ago.
>Did you have miscategorization issues where say there were articles about films underneath cat Gothic architecture and how did you solve this?
Oh, yes.
Then I decided do go through subcategories step by step and control results visual after each 500-1000 records. Nothing more to do.
Now it becomes easier, because there are no huge volumes of new data.
>Is that what you're referring to with I typically perform these checks once a year?
No. There are other tasks and problems. First: Now I support in 10 (or 12) languages. But every wikipage I have already in database could get more synonyms pages in languages that I support - so I need to complete "main article" with synonymes.
Second: duplicates. Sometimes two independent pages (in different languages!) appear about the same object (it is normal). But after month-two-... somebody recognizes, that they are the same and one of them become a synonym for other. I have to recognise such a situation and remove duplicate.
Third: page changes its name (it is important to get resume from wiki).
Some pages disappear. And so on.
Last control for all these things was finished exactly before dataset has been published. So - 4 days ago.
>are you referring to sth else with "Otherwise"?
Not only this.
I wanted more or less valuable description. From wikipedia, not from external link - not from Church web-site about praying-plan for next month :)
Latest comment: 4 days ago1 comment1 person in discussion
Hello
Due to the new wishlist process, this wish which was previously marked with the "long-term opportunity" label will be updated to the "proposed wish" label.
With this new label, the wish comes back to the pool of wishes. It will then be triaged and re-submitted for votes.
