You are on page 1of 24

A TECHNICAL SEMINAR

On
“Dynamic Group Membership in Social
Networks”

Presented by
ARJUN. K. R. 4JN07CS012

Under the Guidance of Coordinators


Mr. Narendra Kumar S. B.E, [M.Tech.] Mrs. Ganavi M. B.E., M.Tech.

Lecturer Lecturer

Mr. Hiriyanna G.S. B.E., M.Tech.


Lecturer
Department of Computer Science and Engineering,
Jawaharlal Nehru National College of Engineering,
Shimoga
Abstract
A social network is a social structure made up of individuals who
are connected by one or more specific types of interdependency. Social
Network developers might want to have users automatically added to or
removed from groups based on properties in their user profiles or other
group membership. The group membership changes over time and the
group division is unrelated to the physical proximity. Grouping the objects
into clusters is done using Clustering algorithm. The application in which
it is at any given time possible to divide its users into groups is considered
and visualized.
Contents
 Introduction
 Clustering
 K-means Clustering Algorithm
 Application
 Conclusion
 References
Introduction
Social Network
A social network represents relationships and flows
between people, groups, organizations

Social Network Analysis


Social network analysis views social relationships in terms
of network theory consisting of nodes and ties.

Advantages
It helps determine a network's usefulness to its
individuals.
Clustering
 Definition - Clustering is the process of
grouping elements into subgroups.

 Examples – Network in a company or


institution.

 Advantages :
 Allow the user to select a group of interest.
 Resolving the problem with node-link
diagrams.
Classification of Clustering

Structure based

Clustering

Content Based
Clustering Techniques
 K-means Clustering.

 C-means Clustering.

 QT Clustering.
K-means Clustering Algorithm
 Choose random number of clusters k.
 Randomly generate k clusters and determine the cluster
centers, or directly generate k random points as cluster
centers.
 Assign each point to the nearest cluster center, using

s (t )i   x j : x j mi( t )  x j m (i t*) for all i *  1,..., k 


 Recompute the new cluster centers.
 Repeat the two previous steps until some convergence
criterion is met (usually that the assignment hasn't
changed).
=

 Recompute the new cluster centers using

1
m ( t 1)
i 
si( t )
x j
x j si( t )

 Repeat the two previous steps until some


convergence criterion is met (usually that the
assignment hasn't changed).
Regarding computational complexity, the k-
means clustering problem is:

 NP-hard in general even for 2 clusters.

 NP-hard for a general number of clusters k even


in the plane.

 If k and d are
fixed, the problem can be exactly
solved in time O(ndk+1 log n), where n is the
number of entities to be clustered, d is dimension
and k is number of clusters.
 Considerclustering of following eight points (with
(x, y) representing locations) into three clusters
(k=3). A1(2, 10) A2(2, 5) A3(8, 4) A4(5, 8)
A5(7, 5) A6(6, 4) A7(1, 2) A8(4,9).

 Randomly choose 3 cluster centers. Initial cluster


centers are: A1(2, 10), A4(5, 8) and A7(1,2).
Initial graph is shown in Fig 1.
Fig 1
(2, 10) (5, 8) (1, 2)
Point Dist Dist Dist Cluster
Mean 1 Mean 2 Mean 3
A1 (2, 10) 0 5 9 1
A2 (2, 5) 5 6 4 3
A3 (8, 4) 12 7 9 2
A4 (5, 8) 5 0 10 2
A5 (7, 5) 10 5 9 2
A6 (6, 4) 10 5 7 2
A7 (1, 2) 9 10 0 3
A8 (4, 9) 3 2 10 2
 New clusters formed are:
 1:{A1}, 2:{A3,A4,A5,A6,A8}, 3:{A2,A7}
 Re-compute the new cluster centers (means) by
taking the mean of all points in each cluster.
Centers of new clusters
C1={2,10}
C2={(8+5+7+6+4)/5, (4+8+5+4+9)/5}={6,6}
and C3={(2+1)/2, (5+2)/2}={1.5, 3.5}
 Above steps are repeated again considering the
new cluster centers. After the second iteration
the clusters formed is shown in Fig 2. New
clusters formed are:
1:{A1,A8}, 2:{A3,A4,A5,A6}, 3:{A2,A7}
Fig 2
 Afterthird iteration the clusters formed is shown
in Fig . New clusters formed are:
1:{A1,A8,A4}, 2:{A3,A5,A6}, 3:{A2,A7}

Fig 3
Applications
 Image Segmentation.

 Visualization of Social Networks.

 Military service.
APPLICATION OF DYNAMIC
GROUP MEMBERSHIP IN
VISUALISING SOCIAL NETWORKS
 Articulated connections between one’s own
“friends” in social networks are obscured, and
can be unearthed only by paging through each
friend’s profile page.

 Node-linkdiagrams are prone in creating a large


mess of nodes and links.
 Most popular areas :
When users dynamically join or leave a community or
conversation.
To find similarities or group users.
To give better visualization(shown in Fig 1).

 Other areas :
Marketing
Defence
Architecture Of Dynamic Group
Membership
Social User Website
Graph Profiles Content and
application
Social
Clustering Relation Networking
algorithm Control Services
Fig 4 – Example of visualizing a social network
Conclusion
The dynamic nature provides flexibility in maintaining
the network cluster around different nodes in the process.
Even with large dataset and many properties, the need of
many users can be answered by the structure and content
based clustering techniques. The clustering technique
reduces the densely populated portion of a node-link graph
by replacing the set of nodes by a single group node. It
helps in visualizing the overall network of organizations by
graphical means. This technique of clustering the users
dynamically and then extracting the information can also be
applied in military services in tracking the terrorists and
criminals who are connected to each other through a
network.
References
[1]Linton C. Freeman, ”The Development Of Social
Network Analysis-A Study In The Sociology Of
Science”,BookSurge,LLC,2004
[2]Jeffrey Heer,Danah Boyd, ”Vizster: Visualizing Online
Social Networks”,2005
[3]Anton Bilos, Technische universiteit Eindhoven,
”Visualization of Social networks”,2007
[4]http://en.wikipedia.org/wiki/Social_network
[5]http://en.wikipedia.org/wiki/K-means_clustering
Thank You

You might also like