Network Systems Flashcards
Egocentric Network
A type of analysis from an individual node/perspective in a network
Whole Network
A type of analysis that describes the network in entirety
Network
A system consisting of a number of similar Nodes where each Node interactive with other certain Nodes in the system
A certain number is agents that are either directly or indirectly connected.
Network Topology
The layout of a network
Determines how different nodes are connected and how they communicate
Path Length
Characteristic of networks
The distance between 2 nodes
Shortest Path Length
The path that connects 2 nodes with the shortest number of edges
Average Path Length
Average number is steps along the shortest paths for all possible pairs of network nodes
Clustering Coefficient
(C) the fraction of associated neighbours of a node that are also connected
How many friends know eachother? (Out of 3)
None: C= 0
Only two: C=1/3
All = C=1
Degree (Branching Factor)
Number of other nodes connected to this node (k)
Degree Distribution
The number of nodes in a network that have a certain degree, or proportion of nodes that have k links
Average degree is 10, if overall average number of branched off a single node are 10
Sparse Network
Characteristic
Has a very small number of connections
Dense Network
Characteristic
Has a very large number of connections
Random Network
Network Type
Generated by random process
Has no particular pattern or regularity in how nodes are connected
K: Low
C: Low
L: Low
Fully Connected Network
Network Type
Each node is connected to all other nodes
K-> N-1
C: High
L: Low
Regular Network
Type
Each node has an identical connection scheme
A clear pattern
K: low
C: high
L: high
Scale-Free Network
Type
Some nodes as as highly connected Hubs and have high degree, but most nodes are of low degree
K: Power Law -> Few nodes with many connections, many nodes with few connections
C: Power Law -> decreases as the node degree increases
L: low
Preferential Attachments
SF
New links are made preferably to hubs
Small World Network
Property of a network
It is very easy to get from one node to another node (ex. Through a hub)
Many clusters of highly interconnected elements (high clustering Coefficient)
Small number of connections between clusters (low characteristic path length)
Linked
Social Network Analysis
When pairs maintain a particular relationship
Ex. Linked by a working relationship
Tied
SNA
Pairs are tied by all relationships they maintain
Ex. Work together but are also drinking buddies
3 Attributes of Relationship
SNA
Content
- characterized by the content of relationship
- depends on what is being studied
- can use many in order to study wider things or if unknown what is most relevant
Direction
- asymmetrical: when info flows in only one direction
- undirected: direction no measured or relevant
Strength
- intensity of relationship
- some only measure if relationship is present or absent (existence of link sufficient to signify important relationship)
- strength included to include weaker links that are still potentially relevant
-
Tie Strength
Depends on number or types of elationships and strength of each individual one
Strength determines likelihood of information being passed
Weak ties impoetant too bcz more likely to he in diff circles and pass on new information
Network Principles Purpose
Relational Properties
- how cohesive a group is
- what subgroups of interconnected actors exist
Positional Properties
- who occupies what position in network
Network Principles
- cohesion (group together according to strong common relationship with eachother)
- structural equivalence (group according to similar relations with others)
- prominence (indicating who is in charge)
- range (in directing context of an actor’s network)
- brokerage (in directing bridging connections to other networks)
Clique
Cohesion
Fully Connected circles
Individually reach eachother directly without going through intermediary or more relaxed def reaching other in 2 steps.
Clusters
Cohesion
Subgroups of highly interconnected actors
Circles/Components
Cohesion
Overlapping cliques show larger netwrok structure
Block Modelling
Pairs that are highly connected (most structurally equivalent) appear together in final results.
Correlation is calculated between all pairs of nodes and then clustered to reorder the cases into sets on basis of their correlated values.
Centrality
Prominence
Measures actors connection in network rather than network as a whole (egocentric)
Highest degree (most lines coming out of node) means it is most central
Star
Prominence
The occupant with the most central position
The hub? Highest access to information and good position to forward to others or prevent it from forwarding
Isolate
Prominence
No connections to others in particular network
Can only recieved info from impersonal sources
May be connected in networks and potentially untapped source of info in this network
Global Centrality
Prominence
Shortest path between actor and every other is calculated and one with shortest distance is the most globally centered
Range
Amount of sources actor has access too.
More ties maintained = more access to sources of info = more access to places to use it
Depends on size of network, number of bridging ties maintained to other networks, and size of network of those who they interact with
Betweenness
Brokerage
Extent an actor sits between two other networks. Can maintain center role without being connected to many others.
Connected to few but is the gatekeeper between main groups and thus controls information.
Preferential Attachment
As network grows, node nodes tent to connect to specific more popular nodes in order to be able to get more connections for themselves.
