A Brief Introduction to
A Brief Introduction to
Social Network
Social Network
Analysis
Analysis
Jennifer Roberts
Jennifer Roberts
Outline
Outline
Description of Social Network Analysis
Description of Social Network Analysis
–
–
Sociocentric
Sociocentric
vs. Egocentric networks
vs. Egocentric networks
Estimating a social network
Estimating a social network
TRANSIMS
What is Social Network Analysis
What is Social Network Analysis
(SNA)?
(SNA)?
An analysis technique which studies
An analysis technique which studies
–
– Relationships between people and groupsRelationships between people and groups –
– How those relationships ariseHow those relationships arise –
– Consequences of the relationshipsConsequences of the relationships
(Christopher McCarty, Director of the UF Survey Research Center) (Christopher McCarty, Director of the UF Survey Research Center)
Nodes = people Nodes = people
Ties = relationships or interactions Ties = relationships or interactions
Two Types of SNA
Two Types of SNA
Egocentric Analysis Egocentric Analysis
–
– Focuses on the individualFocuses on the individual –
– Studies an individualStudies an individual’’s personals personal
network and its affects on that individual
network and its affects on that individual
Sociocentric
Sociocentric AnalysisAnalysis
–
– Focuses on large groups of peopleFocuses on large groups of people –
– Quantifies relationships between people Quantifies relationships between people in a group
in a group
–
– Studies patterns of interactionsStudies patterns of interactions and how these patterns affect
and how these patterns affect
the group as a whole
the group as a whole
From redwing.human.net/~mreed/ warriorshtm/ego.htm
Egocentric SNA
Egocentric SNA
Examines local network structure Examines local network structure Describes the network around a Describes the network around a
single node (the ego) single node (the ego)
–
– Number of other nodes (alters)Number of other nodes (alters) –
– Types of connectionsTypes of connections
Extracts network features Extracts network features
Uses these factors to predict health Uses these factors to predict health
and longevity, economic success, and longevity, economic success,
levels of depression, access to new levels of depression, access to new
opportunities opportunities
From www.stop.hu/ showcikk.php?scid=1005217
Sociocentric
Sociocentric
SNA
SNA
Quantifies relationships Quantifies relationships
and interactions between and interactions between
a group of people a group of people
Studies how interactions, Studies how interactions,
patterns of interactions, patterns of interactions,
and network structure and network structure
affect affect
–
– Concentration of power Concentration of power and resources
and resources
–
– Spread of diseaseSpread of disease –
– Access to new ideasAccess to new ideas –
– Group dynamicsGroup dynamics Raines Laboratory Research Group, U. Wisconsin-Madison
Survey and Interview Data
Survey and Interview Data
Collection Techniques
Collection Techniques
–
– Name generator/name interpreter questionsName generator/name interpreter questions
Questions to elicit list of names: Who do you
Questions to elicit list of names: Who do you
discuss important matters with?
discuss important matters with?
Follow
Follow--up questions: Reports about that persons up questions: Reports about that persons attributes, type of tie, ties between pairs of
attributes, type of tie, ties between pairs of
contacts
contacts
–
– Inflow/outflow of resourcesInflow/outflow of resources
Expensive, subject to human error,
Expensive, subject to human error,
influenced by the nature of the questions
influenced by the nature of the questions
asked
asked
Other Data Sources
Other Data Sources
Indirect measures Indirect measures
–
– Corporate records, event attendance, coCorporate records, event attendance, co--citations, cocitations, co- -authorship, trading patterns, shared affiliations, email,
authorship, trading patterns, shared affiliations, email,
phone calls, computer conferencing
phone calls, computer conferencing
–
– NonNon--invasive, inexpensiveinvasive, inexpensive –
– Not obvious how indirect measures relate to actual Not obvious how indirect measures relate to actual interactions
interactions
Small scale methods Small scale methods
–
– Observation, diariesObservation, diaries
Experimental Experimental
Accuracy
Accuracy
Surveys
Surveys
–
–
accessing validity of people
accessing validity of people
’
’
s
s
reports
reports
–
– Recall and observation do not match up Recall and observation do not match up -- --“
“people do not know, with any acceptable people do not know, with any acceptable accuracy, to whom they talk over any given accuracy, to whom they talk over any given period of time
period of time”” –
– Evidence that people are good at recalling Evidence that people are good at recalling typical interactions, bad at answering about typical interactions, bad at answering about specific time scales
TRANSIM
TRANSIM
–
–
A Method for
A Method for
Estimating Large Social Networks
Estimating Large Social Networks
Assumes transportation network constrains Assumes transportation network constrains
interactions interactions
Creates a synthetic population Creates a synthetic population
Models large
Models large--scale human interactions through scale human interactions through simulations
simulations
Used to study transportation planning, disease Used to study transportation planning, disease
propagation, mobile communications,
propagation, mobile communications, and and demand within the electric
demand within the electric power grid
power grid
Creating a Synthetic Population
Creating a Synthetic Population
Data
Data
–
– Land use and demographic census dataLand use and demographic census data –
– Survey data about daily activitiesSurvey data about daily activities
Information used to create a synthetic
Information used to create a synthetic
population with
population with
–
– List of activities consistent with survey dataList of activities consistent with survey data –
– Access to transportation consistent with Access to transportation consistent with survey data
Activity Planning and Simulation
Activity Planning and Simulation
Creates schedule based on activity lists
Creates schedule based on activity lists
–
– Optimizes sequential plan based on Optimizes sequential plan based on transportation mode
transportation mode –
– Updates schedule based on current Updates schedule based on current congestion levels
congestion levels
Simulation
Simulation
–
– Bipartite graphs contains people nodes and Bipartite graphs contains people nodes and location nodes
location nodes –
– Simulation updates people’Simulation updates people’s locations every s locations every second
Graph Structure
Graph Structure
Set
Set PP of people nodesof people nodes Set
Set LL of location nodesof location nodes Edge
Edge (p, l)(p, l) indicates indicates pp visits visits ll on a normal dayon a normal day
L
L has a power law distribution with has a power law distribution with = 2.8= 2.8
–
– Number nodes with degree Number nodes with degree ii equalsequals
P
P’’ss distribution is concentrated around a small distribution is concentrated around a small average value
average value
Theoretical Model
Theoretical Model
CL
CL
-
-
Model
Model
–
– Generates graphs based on expected degree Generates graphs based on expected degree sequences
sequences –
– Each node is assigned a weight, Each node is assigned a weight, d(ud(u)), equal , equal to its expected degree
to its expected degree –
– Edge probabilities are proportional to node Edge probabilities are proportional to node weights and edge assignments are
weights and edge assignments are independent independent
=
=
⋅
=
σ
σ
Projection
Results
Results
CL,
CL,
fastgen
fastgen
approximation create
approximation create
networks with similar properties to actual
networks with similar properties to actual
data (fig 1)
data (fig 1)
Also include
Also include
–
–
approx algorithm
approx algorithm
description, table, graphs of results,
description, table, graphs of results,
projection graphs
projection graphs