Predicting missing links in complex networks based on common neighbors and distance

Area of Science:

• Network Science
• Data Mining
• Graph Theory

Background:

• Link prediction algorithms based on common neighbors are widely used in complex network analysis.
• Existing methods often struggle with accuracy when predicting links between nodes with few or no common neighbors.
• Network reconstruction accuracy is crucial for understanding network dynamics.

Purpose of the Study:

• To propose a novel algorithm for link prediction that enhances accuracy, particularly for nodes lacking common neighbors.
• To address the limitations of current common neighbors-based algorithms in complex network reconstruction.
• To improve the performance of link prediction in real-world networks.

Main Methods:

• Developed a new algorithm integrating the common neighbors metric with distance-based approaches.
• Evaluated the algorithm's performance on various real-world network datasets.
• Compared the proposed method against existing state-of-the-art link prediction techniques.

Main Results:

• The proposed algorithm significantly improves the accuracy of predicting missing links, especially between nodes with no common neighbors.
• Demonstrated superior performance compared to most existing methods across diverse real-world networks.
• Achieved enhanced accuracy without an increase in computational complexity.

Conclusions:

• The novel common neighbors and distance-based algorithm offers a more robust solution for link prediction in complex networks.
• This method effectively handles challenging cases involving nodes with limited or absent common neighbors.
• The algorithm provides a valuable tool for accurate network reconstruction and analysis.