Individual-centralized seeding strategy for influence maximization in information-limited networks

Area of Science:

• Social Network Analysis
• Information Diffusion Studies
• Behavioral Economics

Background:

• Peer effects drive behavior through interaction networks, amplified by targeting influential individuals.
• Current seeding strategies like the one-hop method face challenges with impractical or expensive network data acquisition.
• Information-limited networks hinder effective targeting of key individuals for interventions.

Purpose of the Study:

• To propose an individual-centralized seeding strategy for identifying influential individuals in information-limited networks.
• To develop a method that infers network structure through respondent nominations, reducing data collection costs.
• To enhance the effectiveness of seeding strategies in real-world field studies and interventions.

Main Methods:

• An individual-centralized approach using follow-up questions (e.g., 'Who has more connections/friends?').
• Constructing a seeding set based on nodes receiving the most nominations from respondents.
• Evaluating the strategy's performance on diverse experimental network datasets.

Main Results:

• The proposed method successfully identifies more influential seeds compared to the one-hop strategy.
• Nomination-based inference acquires significant network information without surveying additional individuals.
• The strategy demonstrates superior performance over existing methods in various network settings.

Conclusions:

• The individual-centralized seeding approach is effective for targeting influential individuals in information-limited networks.
• This method offers a practical and cost-efficient alternative to traditional network surveying for seeding strategies.
• The proposed approach holds potential for improving the efficacy of interventions in real-world scenarios.