jamesmikedup...@googlemail.com wrote: >Sent: 12 March 2010 7:07 AM >To: Al Haraka >Cc: Talk Openstreetmap >Subject: Re: [OSM-talk] [OSM-dev] GSoC'10 > > > >On Fri, Mar 12, 2010 at 5:11 AM, Al Haraka <alhar...@gmail.com> wrote: > > > > It's completely not the "osm way" *as I interpret it* and isn't >going to fly *as long as I am around*. > > > > >I think that is going to far, my point here is not to be negative, but to >contribute something. > >I can imagine that people dont want to force a global model on everyone, >and I agree. > >the current situation of tags being informally defined, partially checked >by various tools and partially supported by many is also not very good.
I'm going to have to wade though the mire here and add my opinion. Why do you argue that the "situation of tags being informally defined, partially checked by various tools and partially supported by many is also not very good". Since the project began I've seen lots of people argue that it just won't work because there is no formal structure, especially for tags, those of us on the other side of the fence argue that's exactly why its been successful. We have built one of the richest geographical data resources available to man and we have done it in a very short time and without any real training whatsoever. So I disagree that the method of freeform tagging is "not very good" and on the contrary it's actually more than surprisingly good. > >if we had at least a set of standardized rules that we could easily apply >to any section of the map, >that would be usable from all programs, in libs and other tools, they we >would move in the right direction. But the dilemma of this is that by standardising the tags we increase the complexity of contribution and its contribution of data that matters wholly and singularly to us. Once we have a "complete" data set then those with a need for a standardised tag system can add those tags that fit strict rules and in fact there is nothing to stop anyone adding this now, they are just additional tags and another way to view the world. > >I can imagine that It would not be difficult to get started with this, >there are efforts already to create rdf representations of the OSM data : Indeed it's actually very easy to come up with standards and many already exist. Just look at GML as one example. But how are you going to train the world of citizen cartographers to use a standard like that, it just can't work. We already have a degree of standardised use via the Map Features type presets used in the various editors, this enables those wanting to add the basics to the dataset to do so without having to think about it. In future tools can be set up to do more of this to remove the need for contributors to think about tags at all, but that in my view is dangerous for two reasons, firstly it reduces the potential of individuals adding data that is of use just because no preset exists and secondly it stops contributors thinking about the meta data at all, its because the contributors have had to think logically about the tags and apply something useful and meaningful that has got us to where we are today, which is, I would argue, the best map in the world. Cheers Andy > >http://linkedgeodata.org/About > >The current ontology is very simple, and contains nothing more than >subclass rules. >http://linkedgeodata.org/vocabulary/ > >Lets look at what rules are important : > >1. Subclassing, a B >they say: a bar is an amenity. ><http://linkedgeodata.org/vocabulary#bar> <http://www.w3.org/2000/01/rdf- >schema#subClassOf> <http://linkedgeodata.org/vocabulary#amenity> . > >2. SubProperties, a name is a label ><http://linkedgeodata.org/vocabulary#name> <http://www.w3.org/2000/01/rdf- >schema#subPropertyOf> <http://www.w3.org/2000/01/rdf-schema#label> . > >3. They have introduced some other things like usagecounts and value >counts, >that is what the the tagwatch should provide. > >So, interesting things that are missing are domain and range. > >Domain : You can say for example that this node is a gas stations and it >has a toilet for wheelchairs. >The property of "wheelchair_access" for example, would be an attribute of >bathroom which is part of a public place, and a gas station is a public >place. > >Or you can use ranges : >Forest is the range of "natural" and a forest has lots of "tree" object. > >Other things would be more advanced, like this disaster area uses these >tags for tagging this and that. Or this sat image is a subset of that >image. We at least could define for example that a certain ontology file is >used to validate this region of the map, you could define them >individually. > >Also there is no one forcing validity of the documents, I am talking about >a simple plugin system that would allow you to start, coupled with a way to >formalize the wiki. > >The task of validating, modelling and using this data is still up to your >user decision. > >I don't see anything more than a better, more customizable josm-validator >plugin. > >In fact, it would be great to see this validator plugin be usable in more >areas, for example as a simple lib that I could plug in to my programs, and >have a command line access to. > >How many different non standard syntaxes to we have for representing rules? >Every OSM tool has to re implement the different rules, has its own syntax >and when something changes lags behind. > >mike > > > >No virus found in this incoming message. >Checked by AVG - www.avg.com >Version: 9.0.733 / Virus Database: 271.1.1/2739 - Release Date: 03/11/10 >21:50:00 > _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
