Database Reference
In-Depth Information
a
b
Fig. 12.6
Social networks in reality mining dataset [
10
].
a
shows the inferred weighted friendship
network, where the weights correspond to the relationship strengths.
b
shows the reported friendship
network
4.4
Semantic Patterns
The social structure of human is one of the fundamental questions in social sci-
ence. As traditional survey methods often suffer from its limited scale, Eagle
and Pentland [
8
,
10
] propose a mobile sensing framework to use human mo-
bility data as indicators of human social network. The Reality Mining project
http://reality.media.mit.edu/
tracked the movement of 94 users for
one academic year and conducted survey about the relationships between those users.
The studies show that human mobility patterns strongly correlate with relationships
among people.
Figure
12.6
shows two social networks of all the participants in the study. Net-
work in Fig.
12.6
a is constructed from mobility data and network in Fig.
12.6
bis
constructed using survey data. These two networks inferred from different data show
similar structure. Such observation provides strong evidence that human movement
data reflects social relationship done by survey. Later in [
9
], Eagle and Pentland fur-
ther propose to use Principle Component Analysis (PCA) to extract representative
behaviors of an individual and of groups.
