Information Technology Reference
In-Depth Information
section, we will test this hypothesis by distinguishing those cities that act as “relays,”
ensuring the articulation of scale for knowledge flows between European national
urban systems. These cities are known as “relay-cities” (
Rozenblat
,
1992
,
2004
).
11.4
Interconnections of European National Urban Systems
To distinguish relay-cities from other European cities, we used the concept of
“articulation nodes” from graph theory (
Berge
,
1958
). In a connected graph, such as
our scientific-collaboration graph of European cities, a node is an articulation node
if the sub-graph obtained by removing it is no longer connected (if removal of the
nodes creates disconnect components). Thus, an articulation node is a bottleneck in
the graph.
In urban geography, articulation nodes can be interpreted as relay-cities that
connect urban systems at different scales: national, European and global. Articu-
lation nodes or relay-cities are crucial vectors for the transmission of knowledge
spillovers within the network because they attract diversified knowledge and ensure
its circulation among the networked European cities. Moreover, the position of such
an intermediary confers a strategic role in controlling the diffusion of knowledge
to other European cities: a relay-city may choose to diffuse knowledge among its
networked cities or to appropriate knowledge for its own profit.
Our analysis shows that few cities are relay-cities (Fig.
11.5
): only 98 cities are
structurally positioned as articulation points in the network. Most of these relay-
cities connect several components: Copenhagen connects 14 components; Paris
connects 11; Athens, Utrecht and Brussels each connect 7; London connects 6; and
Vienna, Berlin, Munich, Dublin and Leipzig each connect 5; . . .
We assume that a given relay-city tends to act as an interface at a particular
scale rather than at all scales (national, European and global). To classify relay-cities
according to the scale at which they act as an interface, we chose a factor-analysis
approach, principal-components analysis (PCA). A PCA can reveal the latent
structure of a set of variables by reducing the attribute space from a large number of
variables to a smaller number of factors.
The rows of the data matrix used for PCA represented relay-cities, while the
columns contained the number of cities with which each relay-city forms a bridge
(i.e., an edge whose deletion increases the number of connected components). These
cities were distributed in three modalities (national, European and global) according
to their relative locations with respect to the relay-city concerned.
Figure
11.6
presents the first two factors or dimensions of the PCA, which explain
77 % of the total variance. Factor 1 (F1) clarifies the scale at which relay-cities act
as interfaces, whereas factor 2 (F2) indicates cities' degree of specialization with
respect to this scale of activity.
The majority of relay-cities tend to ensure interconnection at the European scale.
However, most of them appear to be diversified, ensuring the interconnection of ur-
ban systems at multiple scales. Only Athens and Utrecht act principally as European
Search WWH ::
Custom Search