Graph Data (or Inter-related data)

Node Ranking
Graphs
[ Connectivity Interactivity Similar Properties Belongingness ]

Pairwise Relation Objects

[ Symmetric Asymmetric ]

• Presents networked components in easy-to-comprehend manner

• Represents all forms of information simply with nodes & links
• Facilitates analysis of links to provide complex interpretations
ü Identifying nature of flow of information
ü Revealing pattern of diffusion of data
ü Discovery of shortest paths in network
ü Planning optimal work schedules
ü Highlighting central nodes with prime importance
ü Assisting decision making with critical cost-benefit analysis
• Equipped with software tools to analyze & visualize real/random graphs
Identifying Influence using
Centrality Measures
 Significance of Centrality
• Centrality is a connection-based measuring phenomena to :
ü Recognize important merchants acting as focal point in retail networks

ü Identify most influential personality or event in online social media

ü Detect key strategies in a business-oriented networks
ü Reveal spreaders of news in online social networking communities

ü Find purveyor of goods in supply chain networks

• Centrality can be influenced by :

ü Measure of cohesiveness of connectivity (underlying structure)
ü Nature of transfer of information across network (information flow)

• Several centrality-based measures are available to :

ü Discover influential entities/retailer for efficient spread of business
ü Measure degree of cohesiveness between entities in retail network
 Types of Centrality Measures
• Degree – how well local neighbourhood of a node is packed
• Closeness – how good is degree of reachability of a node from any other
node in networked community
• Decay – rate at which information deteriorates from indirect neighbours
• Betweenness – how well-positioned is a node to act as bridge/gateway
• Information – what impact does absence of a node would result on
network coverage for information spread
• Eigen Vector – what importance a node share in a given community
(measure of popularity/rank)

Note :
- Wide variety of metrics are available to measure the influence of an entity
- No individual measure can be regarded best among all
- Each one of the measures is capturing different aspects of connectivity
depending upon domain/context of application, e.g.: retail, bank, etc.
 Degree Centrality
• Degree of a node captures connectedness of local neighbourhood
(!"#$%& ℎ() !"*&+,%$)
• Normalization factor : , − 1 (maximum possibilities that many exist)

Value 0,1

!0
/0 1 =
,−1
Normalized Version

!3 4
/3 1 = = = 0.2857
15 − 1 14
More is the degree centrality (/0 1 → 1)
!;< 6 ⇒ Better compact neighborhood
/;< 1 = = = 0.4285
15 − 1 14
 Directed Degree Centrality
• Degree centrality in directed graph

"#,%& "#,-./
"#,%& (⃗ = "#,-./ (⃗ =
*−1 *−1 Relatively easier to
reach from node !
• Degree centrality will miss lots of information

It only says :
Ø How large is local neighbourhood
Ø Direct-hop connectivity Difficult to reach
from node ! !
It doesn’t says :
Ø Where a node is positioned in the network
Ø Different ways to reach a node
Ø Whether node is even accessible/reachable from a distance node
 Closeness Centrality
• Closeness Centrality &'( → keeps track of relative distances from node " to
other nodes in the network
• Scales directly with distance, i.e. a node twice as far, is half as central

• Special Case : If node " is connected in (1-hop) with all other nodes in the
network/community , then ∑- '(", $) = 1 ⇒ &'( = 1 (" is ideally central)

Normalization factor
&'01 = 03⁄41
9−1 &'5 = 03⁄54
&'( = &'1 = 03⁄46
∑- '(", $)
&'7 = 03⁄47
Normalized Version &'8 = 03⁄48

! ", $ → Length of shortest path between node " and $

 Closeness Centrality
• An Example
Closeness centrality of node 3

1 14
!"# $ = 14 =
4 + 8 + 12 + 8 32
 Closeness Centrality
• An Example
Closeness centrality of node 3

1 14
./0 1 = 14 =
4 + 8 + 12 + 8 32

1 2 4 8 + 5 7 9 15 + …
12 13 10 6 + 11 14

1×4= 4 2×4 = 8 3×4 = 12 4×2 = 8

Hop Number of Neighbours

 Closeness Centrality
• An Example
Closeness centrality of node 3

1 14
./0 1 = 14 =
4 + 8 + 12 + 8 32

1 2 4 8 + 5 7 9 15 + 12 13 10 6 + 11 14

1×4= 4 2×4 = 8 3×4 = 12 4×2 = 8

Hop Number of Neighbours

 Directed Closeness Centrality
Depending upon the direction , there are two types of closeness :-
• in-closeness
• out-closeness
&*+ // From 6 78 9 … … 9 ← 6
1.!"#,%& (⃗ = ∑
- .(#,0)
&*+
2. !"#,234 (⃗ = ∑ // From 9 78 6 … … 6 ← 9
- .(0,#)

Note :- in-closeness denotes → prestige in a relation-based network

!"#,%& (
8
=
1+1+3+2+1+3+4+2
8
= = 0.47
17
prestige for node 1
 Betweenness Centrality
• Betweenness Centrality : measure of centrality in terms of acting as
crossing point in between routes
• Designates a node to be central if it lies in maximum number of shortest
paths connecting any distinct pair of nodes in the network !.

• Mathematically, betweenness of a node ℎ is expressed as the ratio of

#$ %,'
# %,'
for all pairs of nodes :

-) ., /
() ! = +
%,',) - ., /

-) ., / → number of shortest paths between nodes . and / that contains ℎ

- ., / → denotes total number of shortest paths existent between . and /
 Betweenness Centrality … contd.
• Normalized Betweenness Centrality

*" +, ,
∑&,()"
* +, ,
!" # =
-−1
⟶ any number of nodes taken two at a time
2
 Betweenness Centrality … contd.
• Normalized Betweenness Centrality

*" +, ,
∑&,()"
* +, ,
!" # =
-−1
⟶ any number of nodes taken two at a time
2

Normalization factor : ?
 Betweenness Centrality … contd.
• Normalized Betweenness Centrality

4- 5, 6
∑0,23-
4 5, 6
,- . =
!−1
⟶ any number of nodes taken two at a time
2

!−1 &'(C
Normalization factor : = *
2
&'(
C* = ? ?
 Betweenness Centrality … contd.
• Normalized Betweenness Centrality

70 8, 9
∑3,560
7 8, 9
/0 1 =
!−1
⟶ any number of nodes taken two at a time
2

!−1 &'(C
Normalization factor : = *
2
(&'()! &'( &'* (&'.)! (&'()(&'*)
*!(&'.)!
= *!(&'.)!
= *
 Betweenness Centrality … contd.
• Example 1
1 − 2 = 1/1 = 1
1
1−3=1
5. 6, 7
∑1,34. 2 1−4=1
5 6, 7
-. / = 5 6 1−5=1
"−1
2 2−3=1
3 2−4=1
4
2−5=1
• Normalization factor 3−4=1
3−5=1
(" − 1)(" − 2) 4−5=1
=?
2

• Betweenness Centrality ?
 Betweenness Centrality … contd.
• Example 1
1 − 2 = 1/1 = 1
1
1−3=1
70 8, 9
∑3,560 2 1−4=1
7 8, 9
/0 1 = 5 6 1−5=1
"−1
2 2−3=1
3 2−4=1
4
2−5=1
• Normalization factor 3−4=1
3−5=1
(" − 1)(" − 2) 5×4 4−5=1
= = 10
2 2

• Betweenness Centrality ?
 Betweenness Centrality … contd.
• Example 1
1−2=1
1
1−3=1
5. 6, 7
∑1,34. 2 1−4=1
5 6, 7
-. / = 5 6 1−5=1
"−1
2 2−3=1
2−4=1 = 10
3
4
2−5=1
• Normalization factor 3−4=1
3−5=1
(" − 1)(" − 2) 5×4 4−5=1
= = 10
2 2

• Betweenness Centrality ?
 Betweenness Centrality … contd.
• Example 1
1−2=1
1
1−3=1
60 7, 8
∑2,450 2 1−4=1
6 7, 8
,0 . = 5 6 1−5=1
"−1
2 2−3=1
2−4=1 = 10
3
4
2−5=1
• Normalization factor 3−4=1
3−5=1
(" − 1)(" − 2) 5×4 4−5=1
= = 10
2 2
10
• Betweenness Centrality ,- . = =1
9:;< =><? 10
 Betweenness Centrality … contd.
• Example 2
1 − 2 = 1/1 = 1
1
1−3=1
5. 6, 7
∑1,34. 2 1−4=1
5 6, 7
-. / = 5 6 1−5=1
"−1
2 2−3=1
3 2−4=1
4
2−5=1
• Normalization factor 3−4=1
3−5=1
(" − 1)(" − 2) 5×4 4−5=1
= = 10
2 2

• Betweenness Centrality ?
 Betweenness Centrality … contd.
• Example 2
1−2=1
1
1−3=1
5. 6, 7
∑1,34. 2 1−4=1
5 6, 7
-. / = 5 6 1−5=1
"−1
2 2−3=1
2−4=1 =9
3
4
2−5=1
• Normalization factor 3−4=0
3−5=1
(" − 1)(" − 2) 5×4 4−5=1
= = 10
2 2

• Betweenness Centrality ?
 Betweenness Centrality … contd.
• Example 2
1−2=1
1
1−3=1
82 9, :
∑4,672 2 1−4=1
8 9, :
,2 . = 5 6 1−5=1
"−1
2 2−3=1
2−4=1 =9
3
4
2−5=1
• Normalization factor 3−4=0
3−5=1
(" − 1)(" − 2) 5×4 4−5=1
= = 10
2 2
9
• Betweenness Centrality ,- . = = 0.9
10
 Betweenness Centrality … contd.
• Example 3
1−3=?
1
1−4=1
60 7, 8
∑2,450 2 1−5=1
6 7, 8
,0 . = 5 6 1−6=1
"−1
2 3−4=1
3 3−5=1
4
3−6=1
• Normalization factor 4−5=1
4−6=1
(" − 1)(" − 2) 5×4 5−6=1
= = 10
2 2

• Betweenness Centrality ,- . = ?
 Betweenness Centrality … contd.
• Example 3
1−3=0
1
1−4=0
5/ 6, 7
∑1,34/ 2 1−5=0
5 6, 7
,/ . = 5 6 1−6=0
"−1
2 3−4=0
3−5=0 =0
3
4
3−6=0
• Normalization factor 4−5=0
4−6=0
(" − 1)(" − 2) 5×4 5−6=0
= = 10
2 2
0
• Betweenness Centrality ,- . = =0
10
• Interesting Fact : Border nodes generally will have betweenness as 0
 Betweenness Centrality … contd.
• Example 4

!" #$ = '⁄( = 0
!* #$ = +⁄( = 0.5
!. #$ = $⁄( = 0.667
?
!1 #$ = +⁄( = 0.5
!2 #$ = '⁄( = 0
 Betweenness Centrality … contd.
• Example 4

!" #$ = '⁄( = 0
!* #$ = +⁄( = 0.5
!. #$ = $⁄( = 0.667
? border border

!1 #$ = +⁄( = 0.5 symmetric

!2 #$ = '⁄( = 0
 Betweenness Centrality … contd.
• Example 4

!" #$ = '⁄( = 0
!* #$ = +⁄( = 0.5
!. #$ = $⁄( = 0.667
!1 #$ = +⁄( = 0.5
!2 #$ = '⁄( = 0
 Betweenness Centrality … contd.
• Example 5

!" # = ?
!& # =?
!'& # = 0.522

," # = 0.285
,& # = 0.285 ⟶ comparing with
,'& # = 0.428 degree centrality

Logistic Network
• Interpretation : If someone wants to have a contracts, it would be difficult
without having visited node X=? as an intermediary agent
 Betweenness Centrality … contd.
• Example 5

!" # = 0.103
!) # = 0.255
!,) # = 0.522

-" # = 0.285
-) # = 0.285 ⟶ comparing with
-,) # = 0.428 degree centrality

Logistic Network
• Interpretation : If someone wants to have a contracts, it would be difficult
without having visited node 15 as an intermediary agent