Abstract
Growth and remodelling impact the network topology of complex systems, yet a general theory explaining how new links arise between existing nodes has been lacking, and little is known about the topological properties that facilitate link-prediction. Here we investigate the extent to which the connectivity evolution of a network might be predicted by mere topological features. We show how a link/community-based strategy triggers substantial prediction improvements because it accounts for the singular topology of several real networks organised in multiple local communities - a tendency here named local-community-paradigm (LCP). We observe that LCP networks are mainly formed by weak interactions and characterise heterogeneous and dynamic systems that use self-organisation as a major adaptation strategy. These systems seem designed for global delivery of information and processing via multiple local modules. Conversely, non-LCP networks have steady architectures formed by strong interactions, and seem designed for systems in which information/energy storage is crucial.
Introduction
Link prediction—the task of identifying likely future or missing connections—is a cornerstone of network science. While traditional metrics like "Common Neighbors" assume that shared connections imply future links, these methods often falter in complex, self-organizing biological systems. This paper introduces the Local-Community-Paradigm (LCP) to bridge this gap, identifying the specific topological properties that make certain networks inherently more predictable than others.
The Local-Community-Paradigm (LCP)
The LCP framework posits that the likelihood of a link between two nodes is not just a function of how many neighbors they share, but how well those shared neighbors are connected to each other. In LCP-compliant networks, common neighbors form tight-knit local communities. This structural feature is a hallmark of systems designed for rapid information processing and adaptive self-organization, such as the human brain and protein interactomes.
Mathematical Corrections
This erratum addresses critical typographical errors in the original 2013 publication regarding the normalization of key link-prediction metrics. Accurate implementation of these formulas is essential for researchers utilizing the LCP framework.
- Jaccard Coefficient (JC): The original text incorrectly utilized an intersection (∩) in the denominator. The corrected formula restores the standard definition: the number of common neighbors divided by the union (∪) of all neighbors.
- Cannistraci-Jaccard (CJC): This specialized metric extends the Jaccard index by incorporating the "Common Area" (CAR)—the internal links between common neighbors. The erratum ensures the denominator is correctly defined as a union, properly normalizing the metric for accurate cross-network comparison.
Network Categorization: Processing vs. Storage
The study reveals a fundamental dichotomy in complex network architectures based on their LCP signature:
- LCP Networks: Characterized by high modularity and "weak interactions." These systems (e.g., brain connectomes) prioritize information flow and local processing. Their topology is highly informative, making them excellent candidates for link-prediction algorithms.
- Non-LCP Networks: Characterized by "strong interactions" and stable architectures. These systems (e.g., power grids or transportation networks) are designed for the storage and reliable transport of energy or information. Because they lack the local community structure of LCP networks, topological link prediction is significantly less effective.
Conclusion
The LCP framework provides a robust theoretical lens for understanding network evolution, proving that predictability is a fundamental property of a network's topology. By correcting the mathematical foundations of the JC and CJC coefficients, this erratum ensures that researchers can accurately apply these metrics to distinguish between processing-oriented and storage-oriented systems. Future applications of LCP analysis may offer insights into the breakdown of connectivity in neurodegenerative diseases or aid in the design of more resilient artificial communication networks.
Summary updated