Database Reference
In-Depth Information
In [
5
], the social network structure of YouTube has been analyzed as a first step
toward understanding the kind of media and social space that it represents. The
analysis draws upon a crawl of YouTube's user profiles, characterized by the
aggregate of all tags used by the authors to annotate the uploaded videos. Clusters
of authors and associated keywords were identified through vector-space projection
and hierarchical cluster analysis. Nine author clusters were thus found, corres-
ponding to distinct genres of popular Internet video. Similar clustering was identi-
fied by analyzing the authors' networks of friends, represented as graphs. These
results show that socially coherent activity (friendship among users) is strongly
characterized by semantic coherence (similar descriptive tags).
A different research issue has been addressed in [
16
]. The work is founded on the
observation that Web surfers are not required to register or upload videos in order to
view the existing materials; a large proportion of YouTube's audience is in fact
expected to fall into this category of users. As a consequence, the network among
(registered) users does not necessarily reflect that among the videos. The social
networking among videos has thus been studied, and relationships between pairs of
related videos have been modeled by means of a directed graph. Measurements
on the graph topology revealed definite small-world characteristics. This phenome-
non, also known as six degrees of separation, refers to the principle that items
(e.g., people, files, URL links) within a given environment are linked to all others
by short chains of “acquaintances”. Similar results were achieved on other real-world
user-generated graphs. However, compared with, for example, the graph formed by
URL links in the World Wide Web, the YouTube network of videos exhibits a
much shorter characteristic path length, thus implying a much more closely related
group.
Authors in [
17
] analyze the growth trend in the Flickr social network in order to
understand the link formation process. Flickr can be modeled as a directed
network, where users represent nodes and directed edges model links between a
pair of users. The presence of a link from a user to another does not imply the
presence of the reverse link. As shown in [
17
] the creation of the first link affects
the second, since users tend to rapidly respond to the incoming link by creating a
link in the reverse direction. Since users explore the network by visiting their
neighbors, users tend to connect to nearby users in the network. Furthermore, the
number of links created and received by users is directly proportional to their
current number of links.
2.3.2 Media Content Distribution
The enormous, steady growth of the well-known online media-sharing commu-
nities, such as YouTube, Flickr, Facebook, MySpace, and the emergence of numer-
ous similar Web sites, confirm the mass market interest. While similar on the surface
to standard commercial media distribution systems, UGC collections follow in fact
Search WWH ::
Custom Search