Quantifying External Information in Social Network Analysis: An Application to Comparative Mythology.

Area of Science:

• Network Science
• Computational Social Science
• Information Science

Background:

• Classical social network analysis (SNA) excels at revealing structural relationships but often overlooks domain-specific knowledge.
• Integrating external information into SNA is crucial for a comprehensive understanding of network dynamics and information propagation.

Purpose of the Study:

• To extend classical SNA by incorporating domain-specific knowledge from the network's source.
• To introduce novel metrics: semantic value (centrality measure) and semantic affinity (relationship function).
• To develop a heuristic algorithm for computing semantic affinity based on the shortest capacity problem.

Main Methods:

• Developed an extension of social network analysis to incorporate external, domain-specific information.
• Proposed a new centrality measure (semantic value) and a new fuzzy-like affinity function (semantic affinity).
• Implemented a heuristic algorithm based on the shortest capacity problem for computing semantic affinity.

Main Results:

• Applied the novel methods to analyze Greek, Celtic, and Nordic mythologies, comparing gods and heroes.
• Tested the approach on the Reuters terror news network and a COVID-19 Twitter network.
• Demonstrated that the proposed method yields more meaningful results than existing centrality measures and embedding approaches across all case studies.

Conclusions:

• The novel semantic approach enhances social network analysis by integrating domain knowledge, leading to richer insights.
• The semantic value and semantic affinity provide a more nuanced understanding of relationships within networks.
• This extended SNA framework proves effective and versatile, offering superior analytical capabilities for diverse network types.