Predicting and Correcting Missing Data on Diffusion Processes in Multiplex Networks.

Area of Science:

• Network Science
• Social Network Analysis
• Information Diffusion

Background:

• Diffusion processes are crucial for understanding information spread, innovation adoption, and user behavior in social media.
• Users engage with multiple platforms, necessitating multiplex network analysis for diffusion studies.
• Incomplete data due to privacy and technical limitations hinders accurate network analysis.

Purpose of the Study:

• To investigate the impact of missing infrastructure and diffusion data on network properties.
• To introduce a novel model, MLC-tree, for analyzing diffusion in incomplete multiplex networks.
• To evaluate the effectiveness of the MLC-tree model on synthetic and real-world social network data.

Main Methods:

• Investigated the impact of random and non-random missing infrastructure data on nine diffusion network properties.
• Developed the MLC-tree model based on multiplex diffusion trees for incomplete networks.
• Evaluated the MLC-tree model using both synthetic and real social network datasets.

Main Results:

• Missing data significantly impacts diffusion network properties, including infected nodes, edges, length, and seed nodes.
• The proposed MLC-tree model demonstrates robust performance in incomplete multiplex networks.
• MLC-tree achieved over 50% relative error reduction with 20-80% data missing.

Conclusions:

• The MLC-tree model offers a significant improvement for analyzing diffusion processes in incomplete multiplex social networks.
• Addressing data incompleteness is vital for reliable social network analysis.
• The findings provide a more accurate approach to understanding information diffusion dynamics in complex online environments.