Rui,
Be serious!
I thought that a general meaning of 'topology' would be the basic set
of relations between objects in a real system *that survives* to the
transformation into the set of relations in the representation
(network).
I see no problem in reducing any complex system (including spatial
systems) into a topological network.
In fact, this is the very basic idea of the new science of networks,
in which most of the concern is on the topology, and not on weighted
graphs.
This argument about "topological limitation" is the lowest point of
Ratti's paper and could compete for a prize of "the worse scientific
argument of all times" against the other argument "one of us has been
in Paris" present in the Rejoinder.
About the Wikipedia, I agree 100% with you:
1) the entry about "Spatial Networks" should be rewritten or deleted
2) the entry about "Spatial network analysis software" should be
renamed to "Space Syntax Software".
Send me a small text about Spatial Networks and I do it for you.
Regards!
Lucas
On 26/06/06, Rui Carvalho <[log in to unmask]> wrote:
> space syntax, as a field, relies on spatial elements: axial lines, convex
> spaces or isovists. But SS networks are topological, as the edges (i.e.
> relations between these elements (axial lines, isovists or convex spaces))
> have, so far, been purely topological.
>
> There is a very long tradition in geography to work with spatial networks.
> These are networks where the edges are spatial (see e.g. 'Network Analysis
> in Geography' by Haggett and Chorley, 1969) and network edges are weighted
> by Euclidean distance between the nodes.
>
> Most of Ratti's critique to SS could be rephrased as a statement that SS
> is non spatial, as the edges do not take in consideration euclidean
> distance.
>
> Hillier and Penn reply in the 'Rejoinder to Ratti' p 505:
> "As soon as topological measures of an axial map are weighted by, say,
> length of segment, the integration pattern resulting from configurational
> analysis will always focus on the geometric centre of the system (because
> that is in general metrically closer to all other parts of the system),
> and decrease smoothly from centre to edge. This has two effects. First, it
> means that a short backstreet close to a main centre of the system will
> appear configurationally more `integrated' than a major line remote from
> the geometric centre. Second, it will make the model so sensitive to the
> choice of boundary that it will be this that defines where the centre of
> integration is."
>
> In other words: SS stops working once we shift from topological to spatial
> networks, which is why SS networks were never spatial.
>
> The question then remains: why are those at the helm filling every
> possible web page with "spatial analysis" and "spatial networks"?
>
> Perhaps it's time for more than rhetoric games...
>
>
> BR,
> Rui
>
> _____________________________
> Rui Carvalho
>
> http://www.casa.ucl.ac.uk/people/Rui.htm
> Senior Research Fellow
> Centre for Advanced Spatial Analysis
> 1-19 Torrington Place
> University College London
> Gower Street
> London WC1E 6BT
> United Kingdom
>
--
Lucas Figueiredo
CASA - Centre for Advanced Spatial Analysis
University College London
1-19 Torrington Place
London WC1E 7HB England
E-mail: [log in to unmask]
|