Similarity-based future common neighbors model for link prediction in complex networks.

Area of Science:

• Network Science
• Computer Science
• Data Mining

Background:

• Traditional link prediction methods struggle with evolving networks due to reliance on current common neighbors.
• Existing similarity-based algorithms often lack sufficient accuracy in dynamic network environments.

Purpose of the Study:

• To introduce a novel approach for link prediction by incorporating future common neighbors.
• To evaluate the contribution of future common neighbors to link prediction accuracy.
• To enhance the robustness and performance of link prediction in complex, evolving networks.

Main Methods:

• Definition of future common neighbors.
• Development of the Similarity-based Future Common Neighbors (SFCN) model.
• MATLAB simulations for parameter tuning and comparative analysis.

Main Results:

• Future common neighbors contribute more significantly to link prediction than current common neighbors in complex networks.
• The SFCN model demonstrates higher prediction accuracy compared to eight other algorithms.
• The SFCN model exhibits superior performance robustness across five different networks.

Conclusions:

• The SFCN model effectively leverages future common neighbors for improved link prediction.
• The proposed method offers a more accurate and robust solution for link prediction in evolving networks.
• This research highlights the importance of temporal network information for accurate link prediction.