Abstract
In this paper, we propose a general model for collaboration networks. Depending on a single free parameter "preferential exponent", this model interpolates between networks with a scale-free and an exponential degree distribution. The degree distribution in the present networks can be roughly classified into four patterns, all of which are observed in empirical data. And this model exhibits small-world effect, which means the corresponding networks are of very short average distance and highly large clustering coefficient. More interesting, we find a peak distribution of act-size from empirical data which has not been emphasized before of some collaboration networks. Our model can produce the peak act-size distribution naturally that agrees with the empirical data well.
Introduction
In network science, the Barabási-Albert (BA) model is the gold standard for explaining "scale-free" networks, where a few hubs have many connections while most nodes have few. However, many real-world collaboration networks—such as scientific co-authorship, movie casts, or infrastructure logistics—don't perfectly fit this mold. Some are more exponential, others are "stretched," and almost all exhibit much higher clustering than the BA model predicts. Furthermore, most models treat the size of a collaboration "act" (e.g., the number of authors on a paper) as a fixed number. In reality, act sizes follow a distinct, single-peaked distribution. This paper introduces a microscopic model that explains both the network structure and the distribution of these collaboration sizes simultaneously.
Methods
The proposed growth-based model is governed by two primary parameters:
- Preferential Exponent ($\alpha$): This determines how much an existing node’s current popularity (degree $k$) influences its chance of being picked for a new collaboration. The selection probability is proportional to $k^\alpha$. When $\alpha = 0$, selection is random (leading to exponential distributions); as $\alpha$ increases, the "rich-get-richer" effect intensifies, leading toward power-law distributions.
- $\lambda$: A constant controlling the average number of existing nodes chosen to collaborate with a new node at each step, effectively dictating the average "act-size."
At each time step, a new node enters the system and "collaborates" with a set of existing nodes chosen based on the preferential attachment rule. If those existing nodes haven't worked together before, new edges are created between them, forming a clique that includes the newcomer.
Key Results and Structural Properties
Small-World Effect
The model successfully produces networks with the "small-world" property: very short average path lengths and high clustering coefficients. As shown in Figure 1, the average path length $L(N)$ scales logarithmically with the network size $N$, which is a hallmark of small-world connectivity.
Spectrum of Connectivity and Degree Distribution
By tuning the preferential exponent $\alpha$, the model interpolates across a spectrum of degree distributions. The authors identified four distinct patterns—Exponential, "Arsy-Varsy" (transitional), Semi-Power Law, and Power Law—all of which are found in real-world data. For instance, with appropriate parameters, the model generates a clear power-law distribution, as seen in Figure 5.
Act-Size as an Emergent Property
A significant finding is that the model produces a single-peaked, exponentially decaying distribution for act sizes without needing to hard-code it. This matches empirical observations from scientific collaborations (e.g., Physical Review Letters) where most collaborations involve a few people, and very large groups are rare.
Conclusion
This work provides a robust starting point for simulating social dynamics where the "group" is the fundamental unit of interaction. While the model is currently simplified—omitting weighted edges and assuming static parameters—it captures the essential "small-world" and "scale-free" characteristics of diverse real-world collaboration networks. Future extensions could incorporate hierarchical structures or evolving social preferences.
Summary updated