[HOT] Digital Revolutions Workshop, Bergen University

[Pierre Béland]

#DigiRevCMI New Information Technology Tools in 21st Century Politicshttp://www.cmi.no/news/?1585-digital-revolutions
My presention yesterday was the opportunity to review that last major  OpenSteetMap / HOT Responses in the context of disaster and to show the various management aspects of such interventions plus quality problems / management in the context of such responses.https://twitter.com/pierzen/status/661120335845138432

I presented briefly the Semantic analysis I started of the 2015 Nepal response. Looking at the OSM Planet File for 2015-04-24 (before the Response) and 2015-06-07 (After the response), I measured how the objects are related to OSM features. This important measure of quality, completes other quality measures of OSM data. It also gives us a global measure of quality, and can help us monitor the progression of the crowdsource effort and detect rapidly some tagging problems. The first step is to relate parents / childs (ie. relation, way, node) and find the tags that describe each OSM Feature.

Either before or after the Nepal Response, only 1% of the objects cannot be related to a feature such as highway, building, amenity, etc.  No key / value combination listed on the OSM Map Features wiki page (plus specific HOT disaster keys).  A 1% error shows a high ontologic precision of the data produced.  Data with no feature, is Invisible data. Either, there was syntax error in the key / value, no tag, or a contributor simply added a name or note. We need to look more closely at such patterns and find ways to correct them rapidly.
To show how we can focus on this "Invisible Data" and cure it, I created the JOSM NoFeature Mappaint style. It can be selected from the JOSM Mappaint Preferences. https://josm.openstreetmap.de/wiki/Styles/NoFeature. This Style shows the key-value combinations I selected for my OSM data analysis.  I invite HOT Validators to use this style and test it while validating data.

We also have access to  dynamic data (ie data created, modified, deleted).  I will analyze more in detail and try to identify patterns. Monitoring semantic quality of data produced can help to correct rapidly, revise instructions, etc.

This two day workshop is a great opportunity to discuss with other Digital Humanitarian Network contributors and thanks to Per Aarvik from SBTF and Bergen Universiy who organized this workshop.
 
Pierre 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20151103/825059e0/attachment.html>