OpenStreetMap
Until now, I did not find the time to report on some of the Birds of a Feather sessions in which I participated during the State of the Map in Brussels last month. A Birds of a Feather (BoF) is an informal session which can be filled in by the presenters as they wish. I visited two BoF sessions on Sunday, one about PT-assistent and one about Kartotherian.
PT-assistent
PT-assistent is a JOSM-plugin for the validation and correction of public transport routes. It was developed during a Google Summer of Code project, mentored by Polyglot and implemented by Darya. Hence, it was not surprising that they lead the session. Polyglot is a long time member of the Belgian community and is responsible for a lot of public transport mapping in our country. He started by introducing some of the problems he has with maintaining public transport (bus and tram) routes and the repetitive correction work for routes running over the same OSM ways.
Since someone in the audience was an avid public transport mapper in the Dortmund area, Darya started the demonstration of the plugin in that area. And low and behold, several problems where detected. We tried to find solutions for the problems, but for some one needs to do a local survey. Others were easily fixed, either manually or by the plugin.
The main power of the plugin is that a fix (done manually or automatically) can be repeated for all other routes that have the same problem in the same spot. One comment was that some of the automated fixes was too much “magic”, but even without it, the plugin is great for locating problems.
The plugin only works for public transport routes of version 2. The people interested in the plugin know what this means.
Great tool to put in your belt ! And great presentation as well, thanks Polyglot and Darya.
Kartotherian
The session was lead by Yuri Astrakhan, the lead developer for Kartotherian. This is the framework that is used to deploy all the components needed to make maps for Wikimedia. The session started very technical by describing the software architecture behind their server. I found this very interesting, probably because of my software developer background :-)
There was a lot of interaction with the audience, which included some people that wanted to setup a similar cluster of processes and computers themselves. But it seems that the documentation of the framework is not ready yet for primetime.
At the end of the session many interesting examples where shown using this technology, or better the technology that is configured by Kartotherian. It is great to see what can be accomplished in combination with SPARQL from the Wikidata Query Service. (A nice video tutorial about SPARQL in the Wikidata context by Ewan McAndrew and some examples).
You can check out some examples from the BoF here and here.
What can we, OpenStreetMappers do in the meantime ? Add wikidata tags to objects, because that is the glue between all those components.
Maybe one day, we will be able to combine OverPass and Wikidata SPARQL queries. Imagine what we can do with all that information…
You felt the passion of Yuri for this project in his presentation. And he knows his stuff. It was a pleasure to follow this session.
I do hope others will be inspired by this little write-up and provide some insight in what was presented during the other Birds of a Feather sessions.
Diskuse
Komentář od SK53 z 14. 10. 2016 v 10:37
I recently tried playing with the wikidata for historic/significant trees in the Czech Republic. It seems that the learning curve to get the most out of SPARQL is rather steep: https://twitter.com/AlbinPCLarsson/status/780038828526563328
Komentář od escada z 14. 10. 2016 v 11:14
So far I only tried to find some light towers and city halls in Belgium, to see whether they where already mapped.
You are probably right about the language. I still have the same problem with Overpass, but at least things like “node“man_made”=”lighttower”;” is partially readable unlike “?item wdt:P31 wd:Q39715 .”
Komentář od SK53 z 14. 10. 2016 v 12:36
I found it relatively easy to retrieve the significant trees in the Czech Republic, but quite a lot harder to query things related to their attributes. The SPARQL manual runs to 200 pages which is rather daunting
Komentář od sabas88 z 17. 10. 2016 v 20:17
Here are all the (top 200) trees in a radius of 300 km around Praha (Czech Republic) http://tinyurl.com/jxkwrfa
SELECT ?item ?itemLabel ?location WHERE { ?item wdt:P31 wd:Q10884 . ?item wdt:P625 ?coords . wd:Q1085 wdt:P625 ?centerLoc . SERVICE wikibase:around { ?item wdt:P625 ?location . bd:serviceParam wikibase:center ?centerLoc . bd:serviceParam wikibase:radius “300” . bd:serviceParam wikibase:distance ?dist. } SERVICE wikibase:label { bd:serviceParam wikibase:language “en” } } LIMIT 200
Fast explaination: “wd:” is the prefix of an object (URI in Linked Data terminology) Qxxx is an object id Pxxx is a property id (proposed and defined through a process) Each value could be another object or a string ?variable is a variable (like $x in PHP) which could be used in the SELECT (acting as a column in the result table) or in the WHERE (the ‘filters’) WHERE is where you put the filters: the data model is made of triples (subject predicate object) and this is reflected in the query syntax. each filter is terminated by a dot and subject or object can be both variables.
To use wikidata query service you need a basic understanding of SQL (for how querying works), RDF (for the logic behind the system) and Wikidata to search for properties and objects…
We could see if there’s enough interest to start some sort of discussion / user group :-)