Big Data Applications - Big Data: Related Technologies, Challenges and Future Prospects - page 65

include Twitter, Facebook, and LinkedIn, etc. have been increasingly popular
over the years. Such online social networking services generally include massive
linked data and content data. The linked data is mainly in the form of graphic
structures, describing the communications between two entities. The content data
contains text, image, and other network multimedia data. The rich contents of
such networks bring about both unprecedented challenges and opportunities to data
analysis. In accordance with the data-centered perspective, the existing research on
social networking service contexts can be classified into two categories: link-based
structural analysis and content-based analysis [ 35 ].
The research on link-based structural analysis has always been committed on
link prediction, community discovery, social network evolution, and social influence
analysis, etc. SNS may be visualized as graphs, in which every vertex corresponds
to a user and edges correspond to the correlations among users. Since SNS are
dynamic networks, new vertexes and edges are continually added to the graphs. Link
prediction is to predict the possibility of future connection between two vertexes.
Many technologies can be used for link prediction, e.g., feature-based classification,
probabilistic methods, and Linear Algebra. Feature-based classification is to select
a group of features for a vertex and utilize the existing link information to generate
binary classifiers to predict the future link [ 36 ]. Probabilistic methods aim to build
models for connection probabilities among vertexes in SNS [ 37 ]. Linear Algebra
computes the similarity between two vertexes according to the singular similar
matrix [ 38 ]. A community is represented by a sub-graphic matrix, in which edges
connecting vertexes in the sub-graph feature high density, while the edges between
two sub-graphs feature much lower density [ 39 ].
Many methods against community detection have been proposed and studied,
most of which are topology-based target functions relying on the concept of
capturing community structure. Du et al. utilized the property of overlapping
communities in real life to propose a more effective large-scale SNS community
detection method [ 40 ]. The research on SNS aims to look for a law and deduction
model to interpret network evolution. Some empirical studies found that proximity
bias, geographical limitations, and other factors play important roles in SNS
evolution [ 41 - 43 ], and some generation methods are proposed to assist network
and system design [ 44 ].
Social influence refers to the case when individuals change their behavior under
the influence of others. The strength of social influence depends on the relation
among individuals, network distances, time effect, and characteristics of networks
and individuals, etc. Marketing, advertisement, recommendation, and other applica-
tions can benefit from social influence by qualitatively and quantitatively measuring
the influence of individuals on others [ 45 , 46 ]. Generally, if the proliferation of
contents between SNS are considered, the performance of link-based structural
analysis may be further improved.
Benefited by the revolutionary progress of Web2.0, the use of generated contents
is explosively growing in SNS. SNS is used to generated contents by various
technology, including blogs, micro blogs, opinion mining, photos, video sharing,
social bookmarking, social network sites, social news, and Wiki. Content-based