[Freegis-list] Cooperative mapping project proposal

[Brett Matson]

Hi Adrian and everyone,

Thanks for the great feedback. At the start of this project I had some long
conversions with my thesis supervisor who raised some of the issues you
pointed out. Others points hadn't occurred to me at all. I have outlined
some of my thoughts on the points you made below. These are still just
preliminary ideas that will hopefully provoke more discussion within the
news
group.

Free Software

I will be using free, open source software exclusively throughout the
project.
I agree that the free web GIS software available is very good quality and I
would like to further the free software/spatial data effort that is now very
strong.

I have considered using ESRI shapefiles to store the data because it was
difficult finding a format that could be easily read and written to by free
software. This is why I decided that the Shapelib library, used in
conjunction
with the OpenMap package, would solve this problem. Do you know of any other
data formats or libraries that might be more desirable.

Heterogenous Data Sources

As you mentioned, this topic will require a lot of further consideration.
Metadata will play an important part in attempting to solve this problem.
Could you point me towards some of the metadata efforts you mentioned?


Disparate Dictionaries

This is a good point. It seems that the simplest (but not most elegant)
solution would be to restrict the number of classes available and allow
the users who contribute the data to decide for themselves which class
to use and then they could possibly provide additional information as
metadata if necessary.

I can't see how it would be possible to completely solve this problem
in a system where people are free to add any data they like. Perhaps
standards and guidelines could reduce the problem enough so that the
map is still worthwhile.


Physically

I guess a system where an unrestricted group of people collect data in
their own way will always suffer from these problems. I wonder if simply
attaching metadata that describes the travelling method, speed, position
on the road, etc.. would be enough to cope with this problem?

Accuracy

I like your idea for a data layer to demonstrate accuracy. Based on
the results of the accuracy test, users could have a 'score'
associated with their data providing other users with an indication
of the quality of their data. Users could then improve their score
by upgrading their technology or technique. By adding a temporal
component to each data set, it could be recorded that in 1999 Jill's
data wasn't too good but since November 2000 her data is very accurate.
Users could then vie for the honour of obtaining the highest score
or possibly have a top-ten contributors list.

I don't know how idealistic this would be but its interesting thinking
about the possibilities. Its this kind of user interaction that I hope
would create a strong culture and inspire people to participate and
have fun with it.


A concern that my supervisor and I had was how these heavy restrictions
and requirements would effect the culture of the user group. If
people feel that there is a too strict regime placed on them,
they may be less inclined to participate. An important aspect of this
project is allowing people to use it freely, but as you pointed out, the
map will be worthless if these issues can't be solved in a sensible way.

Thanks again for your help, Adrian. If you have any more ideas I'll be
glad to hear them. I would also like to hear the advice or opinions of
anyone else interested in GIS or GPS. Even an expression of interest in
using the system when it gets up and running would be a good motivator.

A website for the project should be up in the next couple of weeks which
will provide more details.

Cheers

Brett



-----Original Message-----
From: Adrian Custer [mailto:acuster at nature.berkeley.edu]
Sent: Monday, 26 March 2001 3:34 AM
To: Brett home
Cc: freegis-list at intevation.de
Subject: Re: [Freegis-list] Cooperative mapping project proposal


Hello Brett and "Free-GIS list",

Your enthusiasm is to be appreciated so take these comments as response
not criticism. Your goal is great, but there are a huge number of
difficulties with what you are tryng to do.



I won't address the technological difficulties except to encourage you
to use a Free back end to your system. My understanding of web based GIS
is that it is by far the most advance of free software GIS efforts and
that, with a bit of work, you could actually have a whole Free gis
system which is "web based". So, take a leap into the unknown and break
out of the shadow of ESRI. :-)



There are quite a few problems which emerge when future users attempt to
assemble data from heterogenous data sources. Your end vision seems to
be to have a huge amount of data which, although perhaps occaisonally
conflicting, could be amalgamated into a decent gis data base. Most of
the problems I see with the heterogenous origin of the data set are the
problems addressed by the large METAdata efforts in GIS. You message
shows you've thought about these data issues but I would encourage you
to spell out early how to deal with this heterogenous data.


Look from the perspective of your future users comming to get data from
the system. Consider what they are able to say about the data, how they
are able to assess quality and reliability of each data set. Then
consider how they could resolve differences between data sets.

As an example, I have submitted my data set of this tiny place in
ethiopia. You have a back drop of say the Digital Chart of the World and
a quick check shows that my elevations are resonable for that part of
the world so the data is at least plausible. Now another user, has taken
a trip to the area and submits her data. The same base check shows that
the data is plausible so it's accepted. Now a user grabs the two data
sets.

Thematically:

    The data dictionaries will be different. How to decide what we both
    meant by roads. (I had four classes (major, minor, passable,
    walkable) The second user could have had a totally different
    description. Some folks are trying to come up with a standard data
    dictionary but I suspect they will fail because we don't all want
    the same information from the planet. You would have to think how to
    resolve disparate dictionaries (disparate definitions of data
    elements)


Physically

    Did we move down the roads in a similar fashion? Who moved down the
    centerline, who stayed in their lanes? How fast did we move? Or did
    one person survey and another drive?


Accuracy

    I claim to have differentially corrected all my data. How do I prove
    to the next user that my data are accuarate? You should think about
    developing some data layer that would demonstrate accuracy. For
    instance you could require some of the data submitted to be a repeat
    visit to the same location ten days apart. A good metric like this
    would be really helpful. Consider the thematic information:

You could of course punt and simply give the third user all the data and
let them slog throught it making their own decisions. This doesn't help.
You still have to figure out how to provide them with the information
they will need to make these decisions. If you don't your whole system
will just prove useless.

So you are left with imposing some fairly heavy
restrictions/requirements of data submitters for data compliance or
meta-data inclusion or left with doing a lot of resolution, filtering
work on your end. Either is practicable but you are going to have to do
something to ensure some standardization of your data. It's tough and my
personal take on this is that the data could never be well integrated
into such a system unless you started out with that in mind (i.e. we all
gathered data for the GNU atlas in GNU GPS format :-) )

I also think that the quickest way to get a decent global data set is to
force it out of the US goverment. They have some great global data sets
that they are holding on to illegally (e.g. the Digital Chart of the
World Level 1 (or 2?) ). thinking has become mostly that it's better to
force the US government to release the data it gathers into the public
domain as required by law

This page is old but relevant:
http://msnhomepages.talkcity.com/NonProfitBlvd/free_world_maps/index.html

So good luck with your project and do consider some of these issues all
the way through to the end user you have in mind.

cheers,
adrian