Interpersonal Ties and societies

Introduction

Introduction

• We will examine an interesting angle on social networks 
• First we shall go over the basics of social network theory
• Afterwards covering several extensions to this foundation
• Including example networks showing these features

quick recap

• Basics of Social Network Theory is very similar to our prior work
• Associate individuals with nodes, relationships with connections
• Can use regular methods and algorithms for analysis
• Many useful results are known from this i.e Small World Effect
• Problems lie with modelling actual social networks however...

how social networks differ

• Fundamental weakness lies in basic network theory, which is:
• All connections are equivalent in nature
• However, actual relationships are not equivalent!
• Thus we need to find and use extensions to the current base
• We will use three major angles of attack:

Differences between relations

1) Strength

The strength of weak ties

• Paper published by M S. Granovetter in 1973 
• Extremely important paper - over 27.000 citations!
• Introduced concept of 'strong' vs 'weak' ties
• Strong ties - correspond to familial/friendship relations
• Weak ties - acquaintances/co-workers etc

Core argument

• Societies viewed as groups of strong ties (cliques) held together by:
• Weak ties, that connect disparate cliques
• Granovetter claimed these weak ties hold together society as a whole
• Whereas strong ties are responsible for psychological/social wellbeing
• Have important effects on real networks, for example...

Network 1: boston neighbourhoods

• Contrasting fates of two communities in mid-20th century Boston
• West-End of Boston was redeveloped without any attempt at opposition 
• Charlestown however resisted this attempt and was succesful
• Difference lay in number of weak ties and their specific nature

2) Sign

Signed graphs

• Connections are either positive or negative 
• Well-developed mathematical underpinning
• Have many uses in sciences - for example, used to model spin glasses 
• Naturally represent friendly and hostile relations respectively

Untitled Slide

Structual balance theory

• First developed by Harary in 1953
• Core claim is that a given signed graph/triad will be (un)balanced
• Balanced network/triad more stable over time than unbalanced ones
• Can be seen as formalisation of 'the X of my Y is my Z' inferences
• Backed by psychology; direct evidence is limited but supportative

Untitled Slide

Network 2: pardus

• Analysis performed on 300.000 MMORG (Pardus) players 
• Looked at six interaction types across the same network of players
• This includes both positive and negative interactions
• Found that interactions have unexpected, complex correlations
• Overall micro structure mirrors that predicted by SBT

Untitled Slide

3) (A)symmetry

Directed graphs

• Similar to regular graphs, except ties now have direction
• Allows representation of (lack of) reciprocity 
• Again have multiple uses e.g. representing webpage links
• Very powerful method, in combination with tie signs

Untitled Slide

Network 3: fraternity

• Newcomb Fraternity study of 17 new members in 1956
• Study marked subject's attitudes towards each other over one semester 
• Data recorded both asymmetries and time evolution
• Special study as we can measure the dynamic aspect of social networks
• Important as most studies are static - they record one point in time only

Untitled Slide

summary

We have covered:

• Basics of social network theory
• Three core extensions: strength, sign and (a)symmetry
• How these feature in real world networks 

Untitled Slide

Questions?