A stochastic generative model for citation networks among academic papers

Area of Science:

• Bibliometrics
• Network Science
• Computational Social Science

Background:

• Academic citation networks are complex systems.
• Existing models often struggle to capture the nuances of citation dynamics.
• Understanding these dynamics is crucial for analyzing scientific impact and trends.

Purpose of the Study:

• To propose a novel stochastic generative model for directed citation networks.
• To accurately represent the probability of citations based on paper type, importance, and publication time.

Main Methods:

• Developed a generative model incorporating logistic, inverse Gaussian, and exponential/Pareto distributions.
• Used out-degrees for paper type and in-degrees for paper importance.
• Validated the model using Web of Science and arXiv citation data.

Main Results:

• The model successfully generates simulated citation graphs.
• Simulated graphs exhibit similar in- and out-degree distributions and node triangle participation to real data.
• The model demonstrates effectiveness across different scientific domains (e.g., physics).

Conclusions:

• The proposed stochastic generative model offers a more reasonable and appropriate approach to simulating citation networks.
• The model can perform complete simulations independently of original data.
• This work provides a valuable tool for analyzing and understanding the structure of scientific literature.