[HOT] Looking for tagging experts or Humanitarian Workers - Help needed

Andrew Buck andrew.r.buck at gmail.com
Sun May 4 15:27:55 UTC 2014

Hash: SHA1

Hello everyone,

A farily long explanation below, to just see how you can help, just
jump ahead to the end of the email.

I am working on an ontology for OSM concepts and tags.  An ontology is
basically a set of true statements about some domain.  These
statements can then be read by a program called a 'reasoner' which
takes the statements and tries to fill in extra 'inferences' from the
statements to get a more complete picture of a domain.  Finally, you
can then query this model in different ways to answer various questions.

For the OSM domain these statements are things like:

*  "A building is a kind of structure"
*  "An airplane hangar is a kind of building"
*  "A Structure is a kind of real world entity"
*  "A building can be tagged with building=yes"
*  "An airplane hangar can be tagged with aeroway=hangar"
*  "Buildings are of high importance to firefighters"

Once these 4 statements have been entered into the ontology we can
then ask it a question like "What tags are of interest to
firefighters?"  The model would then make the following chain of
inferences to answer the question:

*  "Since a building is of high importance to firefighters and a
building can be tagged with building=yes, then the tag building=yes is
of interest to firefighters."

*  "Since an airplane hangar is also a building, and airplane hangars
can be tagged with aeroway=hangar, then aeroway=hangar is of interest
to firefighters."

*  "Therefore both the tag building=yes and aeroway=hangar are
important to firefighters because either of them could indicate a kind
of building."

Notice that in answering the question the reasoner was able to make
several inferences that we didn't have to manually enter into the
model.  We don't have to manually curate a list of "what is of
interest to a "police officer travelling on foot with a police dog"
because we can simply have lists of things like "what is of interest
to a person on foot" and "what is of interest to a person with a dog",
etc, and then just tell the model how to combine them to make the list
we are actually interested in.

The model can also be used to document things like mispelled versions
of tags, different schemes for tagging the same real world concept,
and even things like tags entered in foreign languages if they exist
in the database.  It will therefore be very useful for things like
designing stylesheets, writing validator rules, creating custom
osm2pgsql schemas, creating routing rules, and so forth.

So what do I need from the community to make this all happen?

Although I can quite easily go through and enter this kind of
information into the model myself, actually figuring out what to enter
is a difficult process once you get past the obvious stuff like roads
and buildings.  What I need is the help of subject matter experts from
many different fields to tell me what is important to them
(humanitarian workers, first responders like police, fire, emt, etc,
cartographers, people doing routing with OSM data, and so on).

I need to know, what kinds of data are of interest to you and how much
you rely on different parts of that data.  I can then program that
into the ontology and then use that information to make it easier for
you to use the OSM data.

To do this I am suggesting we all just meet on mumble as the more
people we get involved in a conversation like this, the more
information ends up coming out of it.  The idea is to find all the
little quirks in the OSM tagging system and the only way to do that is
to get input from all the stakeholders.  So please join mumble
(instructions at the link below if you need them) and join the


I will be anytime during the day (US time) and I figured since today
was a Sunday it might be a good day to get started.

- -AndrewBuck
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/


More information about the HOT mailing list