Link Prediction of Green Patent Cooperation Network Based on Multidimensional Features

Area of Science:

• Innovation Studies
• Network Science
• Environmental Policy

Background:

• Regional green patent cooperation networks are crucial for understanding collaborative innovation.
• Link prediction in these networks can forecast trends and identify partners for technology collaboration.

Purpose of the Study:

• To propose a multidimensional link prediction model for regional green patent cooperation networks.
• To integrate node, path, and content features for improved prediction accuracy.
• To apply the model to the Beijing-Tianjin-Hebei region for practical insights.

Main Methods:

• Developed a model integrating node, path, and content features.
• Utilized the entropy weight method for node similarity indicators.
• Incorporated heterogeneous path analysis and patent text topic analysis for content similarity.
• Employed the Grey Wolf Optimizer (GWO) for optimal weight determination.

Main Results:

• The multidimensional prediction model significantly improves prediction accuracy compared to existing methods.
• Experimental results validate the model's effectiveness in forecasting network evolution.
• The model successfully predicted the green patent cooperation network in the Beijing-Tianjin-Hebei region.

Conclusions:

• The proposed multidimensional link prediction model offers a robust approach for analyzing and forecasting green patent cooperation networks.
• Accurate link prediction aids organizations in strategic partner identification for technological advancement.
• The findings provide valuable insights into regional innovation dynamics and collaboration patterns.