Min HuRebecca Yu
Nayeoung Kim
04/12/23SI 508 Networks 1
Three Questions◦ How do numbers of videos uploaded by users
affect the number of subscriber each user has?◦ Do numbers of videos uploaded by users follow
power law?◦ Do people have the same interests as others if
they are in the same community?
04/12/23SI 508 Networks 2
Picking 10 categories on Youtube◦ Autos & Vehicles, Comedy, Entertainment, Film & Animation,
People & Blogs, Music, News & Politics, Pets & Animals, Sports, and Travel & Events
Look at one community from each Category. (Based on geography location, number of members, common interest, etc)
◦ Number of Videos a User Uploaded◦ Number of Subscribers a User Have◦ Users to Subscriber Network◦ Number of Friends a User Have◦ User to Friend Network◦ Individual and Overall Interest for Videos in
Different Categories.
04/12/23SI 508 Networks 3
Correlation Coefficient: 0.312451 But
04/12/23SI 508 Networks 4
More videos, increasing number of users with a higher number of subscribers
04/12/23SI 508 Networks 5
Correlation Coefficient: 0.550248
04/12/23SI 508 Networks 6
More subscribers, increasing number of friends
04/12/23SI 508 Networks 7
Many Videos Uploaded by Few Users
04/12/23SI 508 Networks 8
Most Videos are Uploaded by Few Users
04/12/23SI 508 Networks 9
Most Subscribers Subscribing to Few Users
04/12/23SI 508 Networks 10
Most Subscribers Subscribing to Few Users
04/12/23SI 508 Networks 11
Most People are Friends to Few Users
04/12/23SI 508 Networks 12
Most Videos Uploaded by Few Users
04/12/23SI 508 Networks 13
Real network has Higher CC Higher centrality measures Lower avg. shortest path
Low CC Avg. shortest path is around 3.5-5.5(exclude
unreachable pairs)
04/12/23SI 508 Networks 15
Real network has Higher CC Higher centrality measures Lower avg. shortest path
# of Vertices: 658
Lower CC Avg. shortest path is around 3.2-4.5(exclude
unreachable pairs)
Real network has Higher CC Higher centrality measures Lower avg. shortest path
# of Vertices: 723
# of Vertices: 4165
The Clustering Coefficient of the network of
member's friends in music group increase from 0.07 to 0.76
Low CC Avg. shortest path is around 3.5-5.5(exclude
unreachable pairs)
04/12/23SI 508 Networks 23
Real network has Higher CC Higher centrality measures Lower avg. shortest path
# of Vertices: 658
Lower CC Avg. shortest path is around 3.2-4.5(exclude
unreachable pairs)
Real network has Higher CC Higher centrality measures Lower avg. shortest path
# of Vertices: 723
# of Vertices: 4165
The Clustering Coefficient of friend network for music group increase from 0.07 to 0.76
The most popular categories overall are Entertainment and Music
If exclude these two categories, users are most interested in the category of their own community
04/12/23SI 508 Networks 31
04/12/23SI 508 Networks 32
33
Pets & Animals41.4%
Pets & Animals41.4% News & PoliticsNews & Politics
68.7%68.7%News & PoliticsNews & Politics
68.7%68.7%
People & BlogsPeople & Blogs12.3%12.3%
People & BlogsPeople & Blogs12.3%12.3%
Travel & EventsTravel & Events12.2%12.2%
Travel & EventsTravel & Events12.2%12.2%
Sports67.2%Sports67.2%
Comedy67.2%
Comedy67.2% Film &
Animation85.5%
Film & Animation
85.5%
Auto & Vehicles85.5%
Auto & Vehicles85.5%
Music82.7%Music82.7%
Entertainment61.6
Entertainment61.6
Identify how people are linked through videos
Find friends’ friend Run more groups on people’s interest and
visualize how people’s interest are connected
04/12/23SI 508 Networks 34
04/12/23SI 508 Networks 35