Abstract:The evolution of many dynamical systems that describe relationships or interactions between objects can be effectively modeled by temporal networks, which are typically represented as a sequence of static network snapshots. In this paper, we introduce a novel random walk based approach that can identify clusters of time-snapshots in which network community structures are stable. This allows to detect significant structural shifts over time, such as the splitting, merging, birth, or death of communities. We also provide a low-dimensional representation of entire snapshots, placing those with similar community structure close to each other in the feature space. To validate our approach, we develop an agent-based algorithm that generates synthetic datasets with the desired characteristic properties, enabling thorough testing and benchmarking. We further demonstrate the effectiveness and broad applicability of our technique by testing it on various social dynamics models and real-world datasets and comparing its performance to several state-of-the-art algorithms. Our findings highlight the strength of our approach to correctly capture and analyze the dynamics of complex systems.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: how to identify the periods when the community structures are stable in temporal networks, and the important changes that these community structures undergo over time. Specifically, the author proposes a new method based on random walks to identify "phases" in temporal networks, that is, the time periods when the network community structures remain relatively stable, and to detect important changes such as the splitting, merging, emerging or disappearing of these communities. ### Main problems and solutions 1. **Problem description**: - **Dynamic changes of community structures**: In temporal networks, community structures will evolve over time, including phenomena such as splitting, merging, emerging or disappearing. - **Challenges of large - scale and noisy data**: When dealing with large and noisy data sets from real - world systems, identifying communities and their evolution remains a difficult problem. 2. **Solutions**: - **Introduction of a new method**: The author introduces a method based on random walks. By analyzing the transition matrices between different snapshots, it captures the similarities and changes of community structures. - **Low - dimensional representation**: This method also provides a low - dimensional representation of the entire snapshot, placing snapshots with similar community structures closer in the feature space. - **Benchmark generator**: To verify the effectiveness of the method, the author develops an agent - based algorithm to generate synthetic data sets with the required characteristics, so as to conduct thorough testing and benchmark evaluation. ### Core steps of the method 1. **Random walk in independent space**: - Perform an independent spatial random walk for each snapshot to obtain the transition matrix \( P_\alpha \). - Calculate the transition probability using the formula: \[ p_\alpha(u, v)=\frac{A_\alpha(u, v)}{d_\alpha(u)}, \quad \text{where} \quad d_\alpha(u)=\sum_{v \in V} A_\alpha(u, v) \] 2. **Snapshot similarity measurement**: - Define the similarity by comparing the transition matrices of different snapshots using the Gaussian kernel function: \[ k(P^\tau_\alpha, P^\tau_\beta)=\exp \left(-\frac{\|P^\tau_\alpha - P^\tau_\beta\|^2_F}{2\sigma^2}\right) \] - Construct a new static network, where the nodes represent snapshots and the edge weights reflect the similarity of community structures. 3. **Temporal random walk**: - Apply a temporal random walk on the new static network to identify the stable phases. - Use spectral clustering to group snapshots into phases with similar community structures. 4. **Low - dimensional embedding**: - Calculate the dominant eigenvector to provide a low - dimensional embedding, so that snapshots in the same phase are close in the feature space. ### Application and verification - **Synthetic data set**: Verify the effectiveness of the method through the generated synthetic data set. - **Real - world data set**: Apply it to various social dynamics models and actual data sets to show its wide applicability and practicality. ### Conclusion The method proposed in this paper can effectively identify the stable periods of community structures and important changes in temporal networks, reducing the computational cost of community detection for each snapshot, and is suitable for the analysis of large - scale temporal networks.

Clustering Time-Snapshots of Temporal Networks: From Synthetic Data to Real-World Applications

Temporal Ordered Clustering in Dynamic Networks: Unsupervised and Semi-supervised Learning Algorithms

Model-Based Clustering of Time-Evolving Networks through Temporal Exponential-Family Random Graph Models

An Efficient Vertex-Driven Temporal Graph Model and Subgraph Clustering Method

Exploring Evolution of Dynamic Networks Via Diachronic Node Embeddings

A framework for spatial-temporal cluster evolution representation and analysis based on graphs

Clustering dynamic networks by discriminating roles of vertices and capturing temporality with subsequent feature projection

Dynamic Network Embeddings for Network Evolution Analysis.

Learning Persistent Community Structures in Dynamic Networks via Topological Data Analysis

Clustering of temporal nodes profiles in dynamic networks of contacts

Evolutionary Robust Clustering Over Time for Temporal Data

Temporal local clustering coefficient uncovers the hidden pattern in temporal networks

Modelling sparsity, heterogeneity, reciprocity and community structure in temporal interaction data

Dynamic Social Network Analysis Using Latent Space Model and an Integrated Clustering Algorithm.

Reconstruction of Stochastic Temporal Networks Through Diffusive Arrival Times

Exploring biological network structure with clustered random networks

Time series clustering via community detection in networks

Generating surrogate temporal networks from mesoscale building blocks

Model‐based clustering of semiparametric temporal exponential‐family random graph models

Clustering of Time-Varying Graphs Based on Temporal Label Smoothness

Dynamic clustering for heterophilic stochastic block models with time-varying node memberships