the influence of social status on consensus building in collaboration networks

http://Learning-Layers-euhttp://Learning-Layers-eu

Learning LayersScaling up Technologies for Informal Learning in SME Clusters

The Influence of Social Status on Consensus Building in Collaboration

Networks

Ilire Hasani-Mavriqi, Florian Geigl, Subhash Chandra Pujari, Elisabeth Lex, Denis Helic

1

Austrian Science Fund: P 24866-N15

http://Learning-Layers-eu

• We tend to create connections and interact with people who have a high social status in our community

• Our behaviour, our opinions are often influenced by actions of such people

• Example: university class – a mentor influences opinions of her student during consensus building

Social Status

2

http://Learning-Layers-eu

Social Status and Consensus Building

• Influence of social status on opinion dynamics is moving from offline to online

• Focus:

– Investigate the influence of social status on dynamical processes that take place in collaboration networks

– Study the interplay between structure, dynamics and exogenous node characteristics and how these complex interactions influence the process of consensus building

3

http://Learning-Layers-eu

Contributions

• Methodologically

– Naming Game (statistical physics) is extended with the Probabilistic Meeting Rule

– Individual differences between nodes in the network are considered

– Through parametrization, explore the emergence and disappearance of social classes in collaboration networks

• Empirically

– Simulate peer interactions in empirical datasets (StackExchange Q&A sites), assuming that the status theory holds and observe the consequences

4

http://Learning-Layers-eu 5

Naming Game Meeting

http://Learning-Layers-eu 6

http://Learning-Layers-eu

Probabilistic Meeting Rule Equation

psl = min (1, e β(ss – sl))

ss – speaker‘s status

sl – listener‘s status

β ≥ 0 – stratification factor (tuning parameter)

7

http://Learning-Layers-eu

The emergence of social classes based on the stratification factor β

8

β = 0, psl is always 1 -> egalitarian society

β = 0.0001, psl decays [0,1] –> ranked society

β = 1, psl is 0 –> stratified society

http://Learning-Layers-eu

Datasets

9

• Datasets from Q&A site StackExchange

• Reputation scores – proxy for social status

• 6 language datasets

#nodes (n), #edges(m), mean (µ), median (µ1/2), standard deviation (σ) of the reputation scores, assortativity coefficient (r)

and modularity (Q)

http://Learning-Layers-eu

Datasets – distribution of reputation scores

10

http://Learning-Layers-eu

Simulations

• The simulation framework is provided as an open source project [1]

• 2 m interactions for the English network, 1 m for other networks

• Investigate various values of the stratification factor β for all networks

• Store the appearance of agents as listeners/speakers, their participation in overall interactions versus successful meetings and the evolution of the agent’s inventory size

• Each agent’s inventory is initialized with a fixed number of three opinions (numbers from 0 to 99)

• These opinions are selected uniformly at random from a bag of opinions to ensure that each opinion occurs with the same probability

[1] https://github.com/floriangeigl/reputation networks

11

http://Learning-Layers-eu

Results

12

Highest convergence rate for 0.0001 < β < 0.0002

http://Learning-Layers-eu 13

http://Learning-Layers-eu

Results

14

http://Learning-Layers-eu 15

http://Learning-Layers-eu

Take Away Messages

• Social status strongly influences the opinion dynamics in a complex and intricate way

• Weakly stratified societies reach consensus at the highest convergence rate, whereas completely stratified societies do not reach consensus at all

• The most important issue in this process is related to low status agents and how their communication is controlled

16

http://Learning-Layers-eu

Future Work

• Engineering consensus building

• Investigate how status and/or network structure can be adjusted to support the process

• Datasets with the strong communities where the consensus reaching is prohibited

17

http://Learning-Layers-eu

Thank you for your attention!

Questions?

Ilire Hasani-Mavriqi

ihasani@know-center.at

Knowledge Technologies Institute, KTI

Graz University of Technology

Social Computing Team, Know-Center (Austria)

18