Prediction and inference in complex networks: a brief review and perspectives

TL;DR

Inference and prediction in complex networks must contend with incomplete and noisy data. The paper surveys network sampling, comparison, link prediction, and time-series–based reconstruction, emphasizing the integration of statistical and machine-learning approaches. It highlights methodological advances, current challenges, and promising directions such as Bayesian methods, graph embeddings, copulas, and causal inference to build principled, scalable frameworks. The work argues that improved uncertainty handling and multiscale, multilayer modeling are essential for impactful applications in science and society.

Abstract

Inference and prediction are fundamental to the study of complex systems, where network data are often incomplete, inaccurate or obtained indirectly. In this paper, we review recent advances in network sampling and comparison, as well as in link prediction and network reconstruction from time series. We summarise key methodological developments and emerging approaches that integrate statistical and machine learning perspectives. We also outline promising research directions for enhancing the inference and prediction of complex networked systems.

Figures (1)

• Figure 1: Key challenges in network inference and prediction. Tasks such as sampling, network comparison and link prediction rely on samples of the network structure. In contrast, approaches such as symbolic regression, statistical dependency analysis and network reconstruction make use of metadata, which is often in the form of time series recorded from nodes and edges.