Patterns of Cascading Behavior in Large Blog Graphs
Blog has become a very popular medium of information because of its ease to access, edit, and publish. Because of its easy accessibility, one blogs is often influenced by other blogs, which means a link in the blog is created for citing. So actually the number of in-links can be seen as a blog’s popularity. And by studying with the number of blogs’ in-links, people can solve questions like “will the popularity of blogs decreases over time” and “what does the patterns of information cascade of blogs look like”.
By aggregating and analyzing 45,000 and approximately 2.2 million blog post, scientists have found some very surprising results.
The first is the number of posts on the blog every day follows a weekly periodicity of which frequency sharply drops off during weekends and resumes in week days. And the popularity of a blog follows, instead of exponentially, a power law whose exponential is -1.5.
The second discovery is the size of distribution of cascades(number of involved posts) also follows a power law with an exponential of -2, and the in-and-out degrees also follows a power law. Also, the pattern of the Post network is very sparse, in which 98% of the blogs are isolated, and the largest connected component has 106,000 nodes but the second largest only has 153.
The third discovery is the shape of information cascades is like “stars” because most of these posts cited a lot of other links but not cited by others.
Source: http://epubs.siam.org/doi/pdf/10.1137/1.9781611972771.60