Understanding tie strength in social networks using a local "bow tie" framework

Area of Science:

• Social network analysis
• Computational social science
• Network science

Background:

• Understanding social tie strength is crucial in diverse settings.
• Modern communication technologies complicate tie strength measurement.
• Existing methods may not fully capture the complexity of social networks.

Purpose of the Study:

• Introduce the social bow tie framework for analyzing social network structures.
• Define metrics to quantify bow tie properties and their relation to tie strength.
• Investigate predictive aspects of the bow tie for tie strength using machine learning.

Main Methods:

• Developed the social bow tie framework, focusing on a central tie and connected actors.
• Defined intuitive metrics to quantify bow tie characteristics.
• Applied machine learning to two distinct social networks (Indian villages, European mobile calls) to predict tie strength.

Main Results:

• Identified key properties of the social bow tie predictive of tie strength.
• Demonstrated that tie strength is influenced by both shared and non-shared connections.
• Revealed a fundamental asymmetry in social interactions based on friend overlap.

Conclusions:

• The social bow tie framework offers a novel approach to understanding social network dynamics.
• Tie strength is not solely determined by mutual connections, but also by unilateral ones.
• This research introduces asymmetry as a critical factor in social tie strength.