Traffic Characteristics and Communication Patterns
in BlogosphereBy Fernando Duarte, Bernardo Mattos, Azer Bestavros, Virgilio
Almeida, Jussara Almeida
Attacked by Ionut Trestian
Introduction – What are we looking at? (I)
Introduction – What are we looking at? (II)
• The paper will try to trick you with big numbers
• 200,000 distinct blogs out of an estimated of 60,000,000 world blogs in 2006.• That’s 0.33%
• Population of Brazil: approx 200,000,000
– That’s 3.33% out of 6,000,000,000
Motivation (I)
• At least the authors don’t give the usual motivation:
- “blogs important in politics, economy etc”
• This would actually be hard in fact of the figures we showed earlier
• Motivation is to study how blogs impact traffic
Motivation (II)
• Traffic engineering• Traffic engineering ???
• Last time I checked ISPs were more worried about clients doing streaming and costly downloads
• Or you want to kill the problem from its source?
Burstiness of access patterns ?
My point?
• The observed burstiness probably comes from external factors – events that trigger users discussion/information sharing
• But the paper never tries to correlate the observed behavior with these facts …
Search engines (I)
Search engines (II)
• Why change search engines by incorporating social information in Page-Rank?
• Search engines and Social Networks have 2 different functions:
– Search engines: finding content
– Social networks: finding what my friends are doing
• Imagine social networks as being bookmarks to your friends
Search engines (III)
• How do people who read get to these different types of blogs, isn’t that important?
– Broadcast probably list links to content
– Parlor is where the discussions take place
– Register ?
Search engines (IV)
• Also the authors never mention what attracted the search engine users to the blog in the first place
– Finding content – rapidshare links
– Getting tricked – spam
• Conclusion
– Terrible idea
– Most search engines ignore blogs anyway
– Makes sense to have both just the same way as most people use both bookmarks and search engines
Distributions (for Gregory)
Conclusions
• Motivation – almost non-existent
• Their Applications even more so …
– Generating blog traffic
– Modifying existing search engines