W3C home > Mailing lists > Public > public-egov-ig@w3.org > August 2012

Re: University of Oxford/ usecase

From: Gannon Dick <gannon_dick@yahoo.com>
Date: Sat, 25 Aug 2012 11:45:37 -0700 (PDT)
Message-ID: <1345920337.71663.YahooMailNeo@web112609.mail.gq1.yahoo.com>
To: "eGov IG \(Public\)" <public-egov-ig@w3.org>
Cc: Tomasz Janowski <twjanowski@gmail.com>, Jeanne M Holm <jeanne.m.holm@jpl.nasa.gov>, "paoladimaio10@googlemail.com" <paoladimaio10@googlemail.com>
Hi Paola,

ref: Truing Up the Sets 

A two "decimal place" alphabetic code trues up the aggregated set while maintaining Member Integrity.  The world+dog totally missed the significance of my "Prime Encoding"[1].  Well, that's not fair, there must be smarter dogs out there I don't know.  Graphs which switch back and forth between Isosceles Triangles and The Normal Distribution (a Gaussian) are can lead to a systemic bias, and inequality.  This happens happens for a combination of two reasons:
1. It happens without any sort of computational warning (Not a number, division by zero, etc.).
2. A set must have an Integer Indexes on the Quartile (1/4, 3/4) and Mean (1/2).  That means there is an Average, and the Average has an Average
The bias is taking the Average of the middle third excluding 1/6th at the high and low ends.
Fairness dictates that 50% are above "average" and 50% are below "average" by some metric, and Society is "naturally" divided into (0-25%,26%-50%,50%-74%,75%-100%) groups.

Unfortunately, here is where I start sounding like a hippy lunatic:

In ISO3166 (2 Character Country Codes)

The representative 0% Member is AA: Unassigned

The representative 25% Member is GM: Gambia

The representative 50% Member is MZ: Mozambique
The representative 75% Member is TM: Turkmenistan
The representative 100% Member is ZZ: Unassigned

In fact, this is not an attempt to identify "World Leaders" but rather the codes were luck of the draw.  There is no (1/3) or (2/3) Member, and for that matter, no multiples of 1/6th.

Tomasz and Jeanne

I'll figure out some better ways to explain, but in the meantime if the "Public Domain" designation is a problem elsewhere (Australia ?). please feel free to suggest more universal licensing terms, and I'll reissue (I may require a note from your dog).


[1] Copyright 2009, me, registration {TXu001614169}.  License: http://creativecommons.org/publicdomain/zero/1.0/
see also: http://www.rustprivacy.org/primeencoding.zip

 From: Paola Di Maio <paola.dimaio@gmail.com>
To: Gannon Dick <gannon_dick@yahoo.com> 
Cc: eGov IG (Public) <public-egov-ig@w3.org> 
Sent: Friday, August 24, 2012 5:52 AM
Subject: Re: University of Oxford/ usecase
Thank you v much Gannon

for picking up the challenge of use cases-

I am not the geekest, but i do have some appreciation for technical beauty

the dataset you cooked up is elegant :-)

btw- is there a way to verify that the aggregated set is
'true'  (with 40 odds count one could check manually but a way to tell
that the meshup is not spitting up false result would be of

Yet  as you probably gather my point is that
socio-technical beauty implies the dataset to correspond to at least
one use case, or similar would like to know question

that we cannot see the use case for the data set does not mean that
there isnt one btw- maybe it will turn out useful in some unexpected
way (alien attack?)





On Tue, Aug 21, 2012 at 10:53 PM, Gannon Dick <gannon_dick@yahoo.com> wrote:
> Seems to me the most pressing problem for Policy Maker Worker Bees is
> translating anecdotal information (mash-ups) into "Normal Distributions" to
> present to their Boss.  The model I've come up with is a blending of
> Mash-ups, Statistics and RDF (Collections).  I hope to make it as easy as
> possible to switch between the vocabularies.  I think that the most pressing
> problem for Policy Makers themselves is the measurement of inequality, and
> for that, you count rooftops because Normal Distributions mix a lot better
> than they un-mix.  To measure inequality, one need only be able to separate
> an "Above Average" Class from a "Below Average" Class, and have the Normal
> Distribution for those Classes.  Mash-ups do this effortlessly, Probability
> Statistics is incapable, and breaks down with uncertainty in the data
> source.
> About the Model ... the geo namespace
> http://www.w3.org/2003/01/geo/wgs84_pos# defines
> altitude    [alt]    The WGS84 altitude of a SpatialThing (decimal meters
> above the local reference ellipsoid).
> All the Organizations, like the University of Oxford in my Model have
> circular view ports (I translate the "Place" to the center of a circle).  A
> Circle is a degenerate Ellipse.  The difference, for counting rooftops, is
> that the use of a circle instead of an ellipse means that you cannot
> abstractly land and look in the side window.  I consider that degenerate :-P
> --Gannon
> ________________________________
> From: Paola Di Maio <paola.dimaio@gmail.com>
> To: Gannon Dick <gannon_dick@yahoo.com>
> Cc: eGov IG (Public) <public-egov-ig@w3.org>
> Sent: Sunday, August 19, 2012 7:50 AM
> Subject: University of Oxford/ usecase
> Hello Gannon
> Thanks a lot for cooking up a recipe to count rooftps, brings a nice
> socio technical view -  and thanks for sharing the method
> from where I come from (the pragmatic world), looks still a bit of a
> theoretical exercise in the sense that I  find it hard to think how to
> use this neat info you pull up
> Could you please help define some use case for this kind of info
> I mean, what can i learn from this view of this data?
> exercise:
> what can I do with the knowledge I gather
> assume
> a)
> b) resident
> c)  public administrator planning for urbanpolicy
> d) any other
> :-)
> P
>> The University of Oxford makes a dandy test bed for Socio-Technical
>> Systems.
>> These are systems which count rooftops and households rather than
>> individuals in a population.
>> There are three types of institutions at Oxford: Colleges, Graduate Only
>> Colleges, and Halls.  Each, 44 in all, have been founded over the last
>> centuries, and the founding dates are available.  No simple RDF List can
>> capture this, however 4 Lists can - three Types + ANY.  The problem, for
>> RDF
>> and SKOS, is the processing of the "first" and "rest".  The model used is
>> the "Standard Model" of Particle Physics - unit sized hard boxes. Much
>> better is a unit sized circle (diameter=2u). This truncates much of the
>> unexplainable combinatorial "fine structure" in the model  which you
>> visualize as gaps in the data.  You see fine structure shadows everywhere.
>> The situation can be compared to a Physician looking at a catalog of
>> thousands of MRI's to find a broken leg when he/she simply needs to look
>> at
>> an MRI of your leg.
>> Anyway, a listing of the data is here[1].  The data base and many
>> supporting files are here[2]. If you load your own, you'll need root
>> privileges to load the Stored Procedures (Space and Time).
>> And last but not least, not to be outdone by the BBC's Olympic Medals and
>> Dr. Brand Niemann
>> ... A mashup of Oxford University Colleges & Halls in time:
>> http://www.rustprivacy.org/2012/roadmap/oxford-university-area-map.pdf
>> When I said I was going to count rooftops, I meant it :-)
>> --Gannon
>> [1]  http://www.rustprivacy.org/2012/roadmap/oxford_sp.txt
>> [2] http://www.rustprivacy.org/2012/roadmap/Oxford.zip
Received on Saturday, 25 August 2012 18:47:07 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:00:48 UTC