Geography Reference
In-Depth Information
registering as user. This place of residence can be easily changed afterwards and
certainly there is no eligibility check. Thus, one might consider our location data
based on user profiles data is a biased and occasionally updated census-type
measurement. After geolocating all individual user data and summing them up on
settlement level, as a secondary step geocoding of settlements had to be done in
order to be able to create network connectivity maps. This was made by locating all
settlement data as points of the centroid of original settlement polygons. The final
settlement centroid points were georeferenced by decimal degree coordinates in
WGS-84 projection.
In each record of the geolocated connectivity database—as mentioned above—
also city to city connections were stored. In other words the dataset contained
information on the location of network nodes (settlements) and edges (number of
connections between settlements). This was registered in the database in forms of
settlement pairs, namely with paired coordinates of the “from node” and “to node”
settlements (network connections were naturally undirected, hence from or to status
does not count and treated as equal). The interconnectedness of nodes then made it
possible to draw network edges as network topology. The database additionally
contained information on the volume of connections between two settlements (the
number of users who have friendships between the two settlements) which data
were applied as weighting attributes of city to city edges.
The nature of OSNs basically assumes that people have online friends often from
the same city (the “from node” and “to node” is identical). In a network database
intra-city connections appear as loops that have to be treated differently from city to
city edges. In our case, therefore, loops were many times left out from connectivity
analyses and mapping.
In our dataset out of the total number of 3,135 Hungarian villages and cities there
were 2,562 settlements, which had active user data. The remaining 573 settlements
did not have iWiW users; majority of these latter locations are very small villages.
Since there were people who did not signify own location precisely, or choose place
of residence outside of Hungary, the number of users in our dataset was somewhat
smaller than the total number of iWiW users. In the analysed database altogether
4,058,505 users were scattered along 2,562 Hungarian settlements. The users have
established 785,841,313 friendship ties in the website, out of which 369,789,373
ties remain within settlement borders (intra-city loops) and 415,653,749 ties are
established between users from two distinct settlements. The network database
covered 1,369,978 settlement-settlement pairs (Table
1
).
Although interesting and useful information are possible to be derived from raw
numbers of intercity ties, in order to deal with size differences of settlements also
log-likelihood ratios were calculated from the raw connectivity data. This measure
is a basic statistical concept; in our case this is the logarithm of the ratio of observed
and randomly expected settlement-to-settlement tie weights (Eq.
1
):
