characterizing online social networks orkut fabricio benevenuto, tiago rodrigues, meeyoung cha,...
Post on 19-Dec-2015
212 views
TRANSCRIPT
Characterizing Online Social Networks Orkut
Fabricio Benevenuto, Tiago Rodrigues, Meeyoung Cha, Virgilio Almedia
Presented by Gregory Peaker, Sankalp Kohli
Dataset• Duration & Time of Data
– Is 12 days between March 26 through April 6 2009 representative?• Mid-terms, Finals, Spring Break?
– Is it representing OSN’s?• “Orkut had the largest number of users and
accounted for nearly 98% of all HTTP requests”
– Accidental clicks– Crawl user profiles April 10-17, 2009– Clickstream data March 26-April 6, 2009
Why Social Network Aggregator?• Why Social Network Aggregator?
– Biased users?– “After discarding non-OSN related log
entries, 17% of HTTP requests related to accessing following four OSNs”
– Is this an actual OSN aggregator?– User Interface of aggregator can
influence user behavior• Is this why Orkut accounts 98%
– Data represents 71% from Brazil and 12% from India
– Section 2.1.2 – what is “single visit”?
Data Limitations• Explain the different OSNs better
– “Linkedin is a network used mainly for professional”
– How about orkut, hi5, myspace?• Session
– We know user is connected, but don’t know if actively using OSN or browsing other tab?
– How they measure sessions?• SessionID on server• Leave tab open, and return in hour• Close tab, and return in 10 min
Flaws in Experiment• Section 5.2
– Disregards online social network information provided on home pages• Wall Posts, Comments, News Feeds
• Section 4.1– “browsing accounted for 92% of all
requests”, what is the other 8% of requests• In same paragraph, “number of users who
ever browsed messages was 13 times larger than those who sent messages”
• Section 4.2 – “MySpace showed a different profile from Orkut”