Community-Contributed Media Collections: Knowledge at Our Fingertips - Community-Built Databases: Research and Development - page 30

In [ 19 ] the analysis performed is more focused on the influence of social contacts
on the bookmarking of favorite photos to estimate the potential spreading capability
of 1,000 favorite photos, over 15,000 unique fans and 35,000 favorite markings.
As shown in [ 19 ] the information dissemination through social links can be
modeled as a slight variation of the model exploited to study the spread of in-
fectious diseases throughout human populations [ 22 ]. A social cascade begins when
the first user includes the photo in his/her list of favorites. Then, the cascade
continues along social links. The following aspects of Flickr in the social behavior
of users have been observed: (a) users maintain their favorite photos indefinitely
(e.g., until photos are removed from the list) and (b) the higher the number of
neighbor users, the higher the dissemination rate. Furthermore, the time required by
the dissemination is in inverse relation to the number of neighbors. Hence, social
links are an effective mechanism for disseminating information in online social
networks.
2.3.3 Semantics of Media Content
In the last few years, an increasing effort has been devoted by the researcher
community to drawing a general picture of the semantic distribution of UGC.
Since this aspect is fundamental to understanding the interests and the predominant
uses, we included it in our proposed taxonomy, shown in Fig. 2.1 .
In [ 16 ], an in-depth measurement study of the statistics of YouTube videos is
provided. Based on a dataset including metadata about more than 3 million videos,
content distribution among categories has first been analyzed, showing that music
videos are prevalent (22.9%), followed by entertainment videos (17.8%) and
comedy (12.1%). Accordingly, the video length distribution indicates that the
majority of videos does not exceed 3-4 min in length.
Similar works have analyzed the distribution of the encyclopedic content available
on Wikipedia. Wikipedia articles are classified into categories and subcategories
exploiting a predefined taxonomy, enriched by article authors. When an article is
uploaded, authors can freely choose one or more categories for their new encyclope-
dia entry, or create a new one. Due to this uncontrolled mechanism, categories
associated with a given article may not always optimally suit its content. The
distribution and the growth of Wikipedia topics were studied in [ 23 ]. The approach
employed for this analysis is based on the evaluation of the semantic relatedness of
each article with taxonomy categories. The topic coverage of Wikipedia article is
unfair. “Culture and arts”, “People”, and “Geography” cover more than 50% of all the
articles, while society and social sciences covers 12%. However, some of the other
topics are rapidly growing. The results of the analysis, reported by the Wikipedia
Foundation, show that the geographic distribution of articles is highly uneven. Most
articles are written about North America, Europe, and East Asia, while only few
about Africa and large parts of the developing world.