tag:blogger.com,1999:blog-2082422783692631078.post5535831372914365943..comments2024-01-08T12:54:17.716+00:00Comments on Maps Matter: On the histories of OpenStreetMap dataSK53-osmhttp://www.blogger.com/profile/05611497805471433828noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-2082422783692631078.post-8109839822993805342011-06-06T15:29:52.085+01:002011-06-06T15:29:52.085+01:00I've added some more images of Berlin Stadt Mi...I've added some more images of Berlin Stadt Mitte on <a href="http://www.flickr.com/photos/sk53_osm/sets/72157626774191915/" rel="nofollow">Flickr</a>. These are month end snapshots from late 2006 to March 2008.SK53-osmhttps://www.blogger.com/profile/05611497805471433828noreply@blogger.comtag:blogger.com,1999:blog-2082422783692631078.post-81888616629018737712011-05-29T19:09:18.532+01:002011-05-29T19:09:18.532+01:00@Steve: I was not trying to suggest that valid tim...@Steve: I was not trying to suggest that valid time histories be added into OSM, certainly not as it stands. I am interested in what approaches and tools are needed for storing vectorised digitised historical data.<br /><br />Broadly I would be in agreement with your point of businesses vs. premises, and, in general, have held off mapping 'POIs' for this reason. Not sure that OSM mappers in general feel the same.<br /><br />The idea of being able to visualise OSM transaction-history, i.e., see what the system was like on a given date, seemed to have a lot of appeal when we discussed in the pub on Tuesday. Most people agreed that it was a struggle to understand exactly what had changed from changeset and object histories.SK53-osmhttps://www.blogger.com/profile/05611497805471433828noreply@blogger.comtag:blogger.com,1999:blog-2082422783692631078.post-25415343040722953822011-05-29T12:12:41.950+01:002011-05-29T12:12:41.950+01:00Steve - I believe there's two dimensions to re...Steve - I believe there's two dimensions to recording historical data. The first one is recording the actual history reflecting changes in the real world. This is to an extent already being done, even through you can't really distinguish real world changes from modifications that were done for other reasons. Mapping history beyond just keeping up with reality is probably not feasible in most places - but in places where there's a lot of OSM activity, I'd say why not? Frankie Roberto started mapping historical buildings in Manchester, see his presentation here http://www.slideshare.net/frankieroberto/mapp-history-on-open-street-map - and as long as it does not get in the way of the core OSM processes I do not see a reason why that would not be a good thing.<br /><br />The second dimension, and you touch on this as well, is retaining historical versions of objects in the database, which makes sense in more ways than you describe. There's not only historical visualization - 'what did the map of London look like on 1-1-2008? - type of applications. A more readily accessible history of contributions to OSM will also make it much easier to answer questions like 'which proportion of contributions in this area will not be carried over after / be affected by the license change?' or simply 'which contributors have been involved in mapping area X / feature Y?'. A quicker answer to these types of questions paves the way for a whole new range of (scientific) analysis of OpenStreetMap data: user contribution patterns, data evolution patterns, quality analysis. <br /><br />Whether improved access to historical OpenStreetMap data should be available through the existing OSM architecture / API is another matter. I can well imagine that this would put too much of a strain on the limited hardware resources, considering the huge amounts of data involved. That should not keep us from moving forward with this though, and the hack weekend that was already mentioned will hopefully help us do that. <br /><br />See also my presentation at the recent WhereCamp EU on this topic. http://www.slideshare.net/mvexel/dealing-with-historical-data-in-openstreetmap - gauging by the amount of discussion it generated, there's quite a bit of interest in it.Martijnhttps://www.blogger.com/profile/03344453706632213587noreply@blogger.comtag:blogger.com,1999:blog-2082422783692631078.post-79467510052662419292011-05-29T03:34:07.927+01:002011-05-29T03:34:07.927+01:00>In OSM we don't make any attempt to discov...>In OSM we don't make any attempt to discover when a pub opened or when it closed, we just know that someone added the pub to the system and that later someone removed it.<br /><br />I'm curious that you think that we should record this information in OSM. IMHO, there are enough difficulties and complexities in OSM's core mission: accurate spatial information about the world at present. The kind of thing you're talking about veers off this path in two dimensions: first, by tracking the business operating inside a premise. We already do this, but I think a dedicated business index would do a much better job, and be much easier to maintain. Secondly, by tracking this kind of history: who defines opening/closure/continuity etc? This kind of thing doesn't need the spatial data, and would be much easier to track elsewhere.<br /><br />There are places where history makes sense, like abandoned railways, rerouted roads, new bridges etc. And in some cases the editing history and actual history will be sort of entwined: there's a pedestrian bridge near me that was destroyed in recent floods. Presuming that it will be rebuilt, we would want the new bridge to be the same way as the old one, which means we should retain the way in the meantime.Stevehttps://www.blogger.com/profile/11963716974666842075noreply@blogger.com