Applications of Graph Analysis

IT5004 Graph Theory 1

 Applications
• Organizational behaviour
• Recommendation & E-Commerce system
• Covert Networks
• Applications to WWW
• Co-citation Network
• Applications to Community
• Collaboration Networks

IT5004 Graph Theory 2

 Organizational Issues
• Organizational Behavior
– study and application of knowledge about how people,
individuals, and groups act within an organization.
– Cooperation and information sharing among the workers is
very important for the success of an organization.
• Graph Analysis can also be used to identify the key or
central persons of an organization which also helps to
understand important to go people in an organization.
• Based on the knowledge obtained from graph analysis,
corrective management strategies can be applied to
improve the efficiency of an organization.
IT5004 Graph Theory 3
 Organizational Issues…
• Team Formation - trust factor and teams cohesiveness is a
central influencer
– centrality and closeness
• Improved information sharing
– sources of information, the structure of information sharing, and
ways to access available knowledge
• Identifying bottlenecks
– un-uniform distribution of workload and resources
• Hidden barriers
– Graph Analysis has been also used as a tool to identify such
hidden barriers, to understand effect of these hidden barriers, and
help people to plan for simple, targeted interventions
IT5004 Graph Theory 4
Recommendation and E-commerce Systems

• Application of Graph Analysis in E-Commerce

recommendations
– Investigate the relationship between different
individuals of system to predict users preference
– Discover the users interaction patterns by
searching users social network and similar likes
and dislikes
– Discover relationships of users to aid in making
recommendations

IT5004 Graph Theory 5

 Recommendation and E-commerce
Systems…
• Graph Analysis in e-learning recommendation
systems
– To present learners with the proper
documentation choice without having sufficient
personal experience or knowledge of available
informatics

IT5004 Graph Theory 6

 Graph metrics for Recommendations
• Graph Analysis makes use of various metrics such as centrality,
cohesiveness, degree of vertex etc. each may reflect different
meaning in recommendation system analysis.
• Example:
– Node with high centrality means it has high impact on other nodes.
– The vertex similarity may be considered as metric to search the
individuals having same interest or preference.
• In this case, system needs to be modeled as a network graph consisting of
customers as nodes and each link represent the same purchase preference.
• The links can be weighted so that the extent of similarity can also be
represented.
• A node with high vertex similarity with existing one is presented with the
items of preference of existing customer.

IT5004 Graph Theory 7

 SNA metrics…
• Cohesiveness property of network
– defines a group of nodes of network bounded with each other by some
relation and may have common characteristics.
– In the context of recommendation systems, a group in transaction network
represents the customers having same interests or preferences.
– Apart from transactional network, customers personal social networks are
also analyzed to predict the groups with similar interests.
– This type of analysis also infers customers preference and further improves
personalized recommendation to the customer.
• Centrality measures
– Generally, customers have a tendency to see and follow what other
influential customers buy.
– Influential customers can be identified by using centrality measures such as
betweenness centrality, closeness centrality and eigen vector centrality etc.

IT5004 Graph Theory 8

Directed
Graphs

(OR)
Individuals
Nodal degree and density for
friendships in ‘Facebook’ applications

Inferences:

1) There are no one isolated (without

friends) in this network (i.e., there
are no individuals with both indegree and
outdegree equal to 0).

2) However, there are two individuals (7

and 9) who did not make any Friend
requests.

3) The mean number of friendship choices

made (and received) is equal to 4.86.

4) The density of the relation is equal to

0.243.
 Different kinds of connectivity in a Social
Media/Networks
• Weakly connected if they are joined by
a semipath
• Unilaterally connected if they are
joined by a path from nj to nj, or a path
from n j to nj
• Strongly connected if there is a path
from nj to nj, and a path from nj to nj;
the path from nj to nj may contain
different nodes and arcs than the path
from nj to nj
• Recursively connected if they are
strongly connected, and the path from
nj to nj uses the same nodes and arcs as
the path from n j to nj, in reverse order
 Prominence: Centrality and Prestige
• An actor to be prominent if the ties of the actor make the
actor particularly visible to the other actors in the network.
• Centrality
– Actor has high involvement in many relations, regardless of
send/receive directionality (volume of activity)
• Prestige
– Actor receives many directed ties, but initiates few relations
(popularity > extensivity)
• Both centrality and prestige indices are examples of measures
of the prominence or importance of the actors in a social
network.
 CENTRALITY
• The three most widely used centrality
measures are:
– Degree
– Closeness
– Betweenness

Node centrality represented in a network. Bigger, redder nodes

have higher centrality values in this representation.
 Degree Centrality
• An actor with high degree centrality maintains
numerous contacts with other network actors.
– Actors have higher centrality to the extent they can
gain access to and/or influence over others.
– A central actor occupies a structural position
(network location) that serves as a source or
conduit for larger volumes of information exchange
and other resource transactions with other actors.
– Central actors are located at or near the center in
network diagrams of social space
 Degree Centrality…
• Actor-level degree centrality is simply each
actor’s number of degrees in a nondirected
graph:
CD(ni) = di(ni)
• To Compare networks of different sizes (g)
– standardize or normalize the degree centrality index
• divide it by the maximum possible indegrees
– (g-1) nodes if everyone is directly connected to i

C’D(ni) = di(ni)/(g-1)
 Degree Centrality…
• Group degree centrality quantifies the dispersion or
variation among individual centralities
– It ranges from 0 to 1, reaching the maximum when
all others choose only one central actor (star) and
the minimum when all actors have identical
centralities (circle)
– Group degree centrality = Σ(cmax - c(ni))/ cmax
• where cmax is the maximum value possible and c(ni) is the degree centrality
of node ni.
Three illustrative networks for the study of
centrality and Prestige
 Closeness Centrality
• Closeness centrality is a useful measure that estimates
how fast the flow of information would be through a
given node to other nodes.
• Closeness centrality measures how short the shortest
paths are from node i to all nodes.
• It is usually expressed as the normalised inverse of the
sum of the topological distances in the graph.
• This sum is also known as the farness of the nodes.
• Sometimes closeness centrality is also expressed
1

simply as the inverse

C (nthe
)  farness.
 
g

C i  d (n ,in )  j
 j1 
IT5004 Graph Theory 18
IT5004 Graph Theory 19
 Betweenness Centrality

• Betweenness centrality measures how often a

node occurs on all shortest paths between
two nodes.
• Hence, the betweenness of a node N is
calculated considering couples of nodes (v1,
v2) and counting the number of shortest paths
linking those two nodes, which pass through
node N.
Example

IT5004 Graph Theory 21

 Betweenness Centrality

• A central actor occupies a “between” position

on the geodesics connecting many pairs of
other actors in the network.
• Actor betweenness centrality for actor i is the
sum of the proportions, for all pairs of actors j
and k, in which actor i is involved in a pair’s
g jk (n i )
geodesic(s) CB (n i )   g
jk jk
 Example
• In the star graph, actor n1 has betweenness = 1.0
while the six peripheral actors = 0.0.
• All circle graph actors have the same betweenness
(0.2).
• In the chain graph, the two end actors have no
betweenness (0.0)
– The exactly middle actor n1 has the highest
betweenness (0.60), while the two adjacent to it
are only slightly less central (0.53).
 PRESTIGE
 Prestige measures of prominence
 Applicable only to directed graphs
 the differences between sending and receiving relations
 Degree Prestige
 A prestigious actor enjoys high popularity, shown by receiving many ties from others.
 Hence, measure actor-level degree prestige as indegree
 To standardize the degree prestige index, divide indegree by network size (g-1).
 Proximity Prestige
 This analog to closeness centrality considers the proximity of actor i to other actors in its influence
domain, the set of all network actors that can reach actor i, directly and indirectly.
 Status or Rank Prestige
 “It’s not what you know but whom you know” – or, more specifically, your own prestige is a function of
the prestige ranks occupied by the alters in your ego network.
 An actor chosen by many high-ranked others acquires higher prestige than an actor that is the target
of only low-ranked choosers.
 One proximity measure is average distance from influence domain actors j to
actor i: d(n j , n i )
 Ii
IT5004 Graph Theory 25
 Covert Networks
• The covert networks are hidden
– the actors of such network does not disclose their
information to the external world.
– Eg. The terrorist and criminal networks
• Uses of Graph Analysis [Application]
– to understand the communication and structure
of terrorist cells
– to predict terrorism activities
• Eg. recruitment, evolution, and ideas diffusion

IT5004 Graph Theory 26

• Graph Analysis techniques applied to terrorists network varies from basic
measures to complex graph algorithms and data mining techniques.
– The basic measures includes betweenness and degree centrality measures, cohesion
factors, closeness etc.
– Also, the network structure or topological measures are required in this type of analysis
which may include node neighborhood search.
• Graph Analysis considers terrorists networks analysis as a problem of
connecting dots - Connecting multiple pairs of dots exposes the total
network.
– Centrality widely used measure to identify key players in terrorist network and further
the actors linked to these key players can be detected to reveal the whole network.
– To facilitate this, the regular day-to-day activities of the key players are monitored.
– The hidden actors are discovered by monitoring contact and the extend of contacts of
known terrorists with other people.
– In order to measure the location of actors in the network, various measures such as
centrality may give us insight into the details of connector actors, mavens, leaders,
bridge actors, isolates, clusters, core actor, peripheral actors etc.
– Structural cohesion is also used to find connectors among group of actors.
– This measure is used to identify sub-groups in an organization having similar features,
skills and involvement in particular event.
– With the advanced graph theoretic and link analysis techniques, graph analytics is
applied to terrorists network to persecution of criminal activities.
IT5004 Graph Theory 27
 Web Applications
• World Wide Web as a social network
– analyzing the navigation and usage patterns of web, better
improvements in web algorithms for gathering, searching and
discovering information can be achieved
– the evolution of web structure can be predicted
– discover community
• discover the individuals community and their interest in particular knowledge
domain.
– Multi-mode network can be created by representing web resources,
persons and community as nodes and web resources being used by
community as edges to discover group of likely or similar persons by
finding cliques with respect to each knowledge domain.
– Graph Analysis is also used in search engines such as google to
enhance keyword search quality.
IT5004 Graph Theory 28
 Community Welfare
• Graph Analysis is used to analyze different
types of relations such as:
– communication patterns, physical contacts, etc
– may reveal the patterns of human contact which
may lead to spread of disease
– Mass surveillance
– Knowledge sharing and enhancement

IT5004 Graph Theory 29

 Communities
• Community: “subsets of actors among whom there are
relatively strong, direct, intense, frequent or positive ties.”
-- Wasserman and Faust, Social Network Analysis, Methods and Applications
• Community is a set of actors interacting with each other
frequently
– e.g. people attending this conference
• A set of people without interaction is NOT a community
– e.g. people waiting for a bus at station but don’t talk to each other

• People form communities in Social Media

 Example of Communities
Communities from Communities from
Facebook Flickr
 Community Detection
• Community Detection: “formalize the strong social groups
based on the social network properties”
• Some social media sites allow people to join groups, is it
necessary to extract groups based on network topology?
– Not all sites provide community platform
– Not all people join groups
• Network interaction provides rich information about the
relationship between users
– Groups are implicitly formed
– Can complement other kinds of information
– Help network visualization and navigation
– Provide basic information for other tasks
 Subjectivity of Community Definition
Each component is a
A densely-knit community
community

Definition of a community
can be subjective.
 Taxonomy of Community Criteria
• Criteria vary depending on the tasks
• Roughly, community detection methods can be divided into
4 categories (not exclusive):
• Node-Centric Community
– Each node in a group satisfies certain properties
• Group-Centric Community
– Consider the connections within a group as a whole. The group has
to satisfy certain properties without zooming into node-level
• Network-Centric Community
– Partition the whole network into several disjoint sets
• Hierarchy-Centric Community
– Construct a hierarchical structure of communities
 Collaboration Networks
• Collaboration network consists groups of
persons working together to perform
particular activity
• The widely studied collaboration network by
researchers in context of SNs are Science Co-
authorship collaboration network and movie
actor collaboration networks

IT5004 Graph Theory 35

The illustration of collaboration network between research institutions

IT5004 Graph Theory 36

 Co-authorship network
• To study dynamics in patterns of interactions
between educational entities or communities
– helps for strategic planning of research and development
• Since, scope of research subject and persons subject
interest may vary over time, these networks has
been viewed as a dynamic social networks
• The required datasets for co-authorship network
analysis is mostly extracted from sources including
scientific journals, bibliographic records and digital
libraries.
IT5004 Graph Theory 37
A co-authorship network extracted from DBLP. Bigger nodes are cluster-nodes.
IT5004 Graph Theory 38
 Co-authorship network…
• The observations and results of time series and location
based analysis captures the nature and characteristics of
research subject over time and location.
– This helps to identify the scope of research discipline at
particular location so that further new inventions in same can be
promoted at respective region and using skills of subject experts.
• The Co-authorship network analysis also helps to study and
understand the interdisciplinary research which is key
factor for innovation
• Other examples for collaborative networks: Movie actor
network and knowledge collaboration network

IT5004 Graph Theory 39

 Graph Analysis metrics
• The cohesion is used to identify the subgroups within network with
respect to each research subject.
• The node similarity measure in this context represents extend of
similar subject skills.
– This will help to identify group of persons to engage in particular research
knowing few expertise in that area.
– The identified hub in sub-network can represent the key researcher in
that sub-network.
• Density
– The scope and popularity of particular subject in its evolution is measured
by computing density over time and location.
• The various centrality measures are also used to analyze the impact
of collaboration of researchers on research in particular discipline

IT5004 Graph Theory 40

 Co-Citation Networks
• The Co-citation network can be viewed as a bipartite
graph showing linkage between two different groups
of documents.
• Basic application of co-citation analysis is to study the
scientific communication.
– Co-citation is used as a measure of similarity between two
objects.
– Co-citation analysis helps to understand the status and
structure of scientific research.
– Basic two approaches of co-citation are author co-citation
and document co-citation.
IT5004 Graph Theory 41
A journal co-citation network. Each node of the network represents a journal and
more two journals are cited together, the closer the relationships between them.

IT5004 Graph Theory 42

 Graph Analysis Metrics
• The node similarity measure is used to find
similarity between two articles or publications.
– In this case, the nodes represent papers and existence
of link shows that two articles were cited in other
articles.
• Centrality represents the scope or importance of
paper or respective research subject.
• While analyzing the relation between different
disciplines, the cohesion property is used to
observe how close two subjects are.
IT5004 Graph Theory 43