Relative-path-based algorithm for link prediction on complex networks using a basic similarity factor.

Area of Science:

• Network Science
• Graph Theory
• Data Mining

Background:

• Complex networks are ubiquitous, with link prediction crucial for understanding network evolution and reconstruction.
• Current similarity-based methods often overlook path information between nodes and their neighbors.

Purpose of the Study:

• To propose a novel link prediction method that incorporates path information between nodes and their neighbors.
• To enhance the accuracy and robustness of link prediction in complex networks.

Main Methods:

• Developed a relative-path-based method for link prediction.
• Utilized factor information from paths between nodes and neighbors, in addition to paths between node pairs.
• Addressed parameter determination for the proposed and existing algorithms.

Main Results:

• The relative-path-based method demonstrated superior prediction accuracy compared to other existing methods.
• Experimental results on six diverse real-world networks validated the method's effectiveness.
• The proposed method exhibited robust performance across different network types.

Conclusions:

• Path information between nodes and their neighbors is a valuable feature for link prediction.
• The relative-path-based method offers a significant advancement in link prediction accuracy and robustness.
• This approach contributes to more accurate network reconstruction and the study of network dynamics.