W. Zhang, B. Derudder, J. Wang, F. Witlox / Bulletin of Geography. Socio-economic Series / 31 (2016): 145–160
156
Fig. 5. The rank-size distributions of cities’ degree centralities in the original network and the transformed network
Source: Own studies
5. A benchmark test using the data
on Weibo users’ inter-city movements
In order to evaluate the validity of our method, we
need compare the transformed network to a measure
of actual flows of people. Due to the difficulty of
finding a corresponding database of flows of HSR
passengers, here we utilize a database of Weibo
4
-us-
ers’ inter-city movements, which represents a specif-
ic part of tangible flows of people. It can be argued
that the flows of HSR passengers and Weibo-users’
inter-city movements have similar characteristics.
The reason is that they serve relatively similar user
markets: the market of HSR is mainly oriented to
business travel and leisure tourism of citizens with
certain economic means (Wu et al., 2013; also see
Zheng and Kahn, 2013: ‘poor rural migrants would
not choose bullet trains’); and most of social media
users are young adults who have certain econom-
ic capacities that include the use of smartphones.
Both the collection and subsequent transformation
of Weibo data follow the methodology developed in
Zhang et al. (2015); here we summarize the main
tenets.
Similar to other social media services (such as
Facebook and Gowalla), Weibo users are allowed
to share their location through a mobile application
that is commonly known as a geo-tagged server, thus
generating massive location records contributed
by millions users. We transform the geo-tagged
information into individual travelling trajectories
by connecting users’ registered place and their
visited cities. In practice, we employ an Application
- 10.1515/bog-2016-0010
Downloaded from De Gruyter Online at 09/23/2016 11:26:34AM
via Universiteit Antwerpen
W. Zhang, B. Derudder, J. Wang, F. Witlox / Bulletin of Geography. Socio-economic Series / 31 (2016): 145–160
157
Programming Interface (API)
provided by Weibo
to crawl all Weibo-users’ travel records submitted
within the YRD region from March to August 2014.
This dataset contains 3 million inter-city footprints;
each record represents a directional inter-city flow
of a person. Finally, the directional network of
Weibo users’ flows was converted to an undirected
one by combining opposite directional links.
By means of a Pearson correlation measure, we
first compare the similarity between both networks
(the transformed network and the original network)
and the benchmark network of Weibo-users’ inter-
city movements in terms of cities’ connectivities.
The correlation coefficients show that, in gener-
al, there is a more similar relationship between the
transformed network and the benchmark network
(r = 0.87 at the 0.001 significance level) which ex-
ceeds the coefficient for the original network
(r = 0.76 at the 0.001 significance level). We also
plot the rank-size distribution of cities’ degree cen-
tralities in the Weibo-users’ movements network
(fig. 5) to compare networks’ structural similarity.
An intuitive sense is that the curve of the Weibo-
users’ network is closer with the curve of the
transformed network. We mathematically compute
the flattening ratio of the Weibo-users’ network
(0.29), which is indeed closer with the flattening
ratio (0.23) of the transformed network.
6. Conclusion
The purpose of this paper has above all been
methodological: we propose to rethink some of the
discrepancies between physical infrastructure net-
works and actual flows occurring in these networks,
focusing on the lens of the railway system. We did
so by (1) assessing some limitations in commonly
used measures of inter-city rail connections and (2)
setting out an alternative approach to approximat-
ing passenger flows in railway networks.
Previous empirical research on measuring
inter-city linkages through the lens of railway
infrastructures has tended to use proxy strategies,
where (1) measuring the potential for interactions
by train and (2) measuring the volume of trains
making inter-city connections stand out as the
two main strands due to the lack of data on actual
passenger flows. However, the method of measuring
the potential for interactions only mirrors the quality
or efficiency of train transport infrastructures itself
rather than considering the ‘direct demand’ for inter-
city linkages. And, the proxy of using the volume of
trains structurally predetermines a flatter structure
in the urban hierarchy than warranted.
This research has shown that ‘dwell time’ at
train stations reflects the length of the alighting
and boarding process, and we use this insight to
estimate actual interaction through the application
of a bimodal network projection function. The em-
pirical application to the high-speed railway (HSR)
network within the Yangtze River Delta (YRD) re-
gion revealed that the transformed network varies
from the original network to a large extent: (i) the
positions of transit cities along main transport cor-
ridors in the YRD urban system are drove down
while some arguable sub-central cities stand out; (ii)
inter-city connectivities tend to be more hierarchi-
cal; and (iii) the flattening ratio has decreased to
60% of the original flattening ratio. Moreover, the
validity of our method has been evaluated through
a comparative analysis with Weibo-users’ inter-city
movements, verifying that the transformed network
more parallels tangible flows of people.
We believe our paper makes two contributions
to the literature. The first is to remind researchers
to re-examine the validity of proxy strategies when
measuring inter-city transport flows. With the ex-
ception of recent research on airline networks in
the context of the world city literature, relative-
ly limited attention has been paid to the degree to
which these infrastructure networks reflect the ac-
tual flows they undergird. In this regard, this article
offers empirical evidence for the structural deter-
minism of using train networks per se, as these
tend to result in flatter networks. Second, the cen-
tral contribution of this paper has been to set out
an alternative method of approximating actual flows
in railway networks, which permits practical appli-
cations in simulating flows of railway passengers
in other cases.
Apart from empirical applications in other cases,
further research issues also include: discussing other
modes of constructing equations, discovering alter-
native perspectives to approximating actual flows in
railway networks, investigating the biases between
the infrastructure provision and corporeal flows in
- 10.1515/bog-2016-0010
Downloaded from De Gruyter Online at 09/23/2016 11:26:34AM
via Universiteit Antwerpen