Static Network Representations Research Articles

Contrastive representation learning on dynamic networks

Representation learning for dynamic networks is designed to learn the low-dimensional embeddings of nodes that can well preserve the snapshot structure, properties and temporal evolution of dynamic networks. However, current dynamic network representation learning methods tend to focus on estimating or generating observed snapshot structures, paying excessive attention to network details, and disregarding distinctions between snapshots with larger time intervals, resulting in less robustness for sparse or noisy networks. To alleviate these challenges, this paper proposes a contrastive mechanism for temporal representation learning on dynamic networks, inspired by the success of contrastive learning in visual and static network representation learning. This paper proposes a novel Dynamic Network Contrastive representation Learning (DNCL) model. Specifically, contrast objective functions are constructed using intra-snapshot and inter-snapshot contrasts to capture the network topology, node feature information, and network evolution information, respectively. Rather than estimating or generating ground-truth network features, the proposed approach maximizes mutual information between nodes from different time steps and views generated. The experimental results of link prediction, node classification, and clustering on several real-world and synthetic networks demonstrate the superiority of DNCL over state-of-the-art methods, indicating the effectiveness of the proposed approach for dynamic network representation learning.

Locating community smells in software development processes using higher-order network centralities

Community smells are negative patterns in software development teams’ interactions that impede their ability to successfully create software. Examples are team members working in isolation, lack of communication and collaboration across departments or sub-teams, or areas of the codebase where only a few team members can work on. Current approaches aim to detect community smells by analysing static network representations of software teams’ interaction structures. In doing so, they are insufficient to locate community smells within development processes. Extending beyond the capabilities of traditional social network analysis, we show that higher-order network models provide a robust means of revealing such hidden patterns and complex relationships. To this end, we develop a set of centrality measures based on the MOGen higher-order network model and show their effectiveness in predicting influential nodes using five empirical datasets. We then employ these measures for a comprehensive analysis of a product team at the German IT security company genua GmbH, showcasing our method’s success in identifying and locating community smells. Specifically, we uncover critical community smells in two areas of the team’s development process. Semi-structured interviews with five team members validate our findings: while the team was aware of one community smell and employed measures to address it, it was not aware of the second. This highlights the potential of our approach as a robust tool for identifying and addressing community smells in software development teams. More generally, our work contributes to the social network analysis field with a powerful set of higher-order network centralities that effectively capture community dynamics and indirect relationships.

Navigability assessment of large-scale redesigns in nine public transport networks: Open timetable data approach

Redesigns of public transport networks are complex undertakings requiring a careful balancing of several conflicting perspectives (e.g., user requirements vs system performance) and aims (e.g., increasing spatial coverage, increasing frequency). Current assessment tools omit an explicit focus on navigability, often identified as a key aspect of the user perspective. For understanding the multidimensional perspective of navigability, this research introduces an assessment framework with both system and journey-level measures. The system-level measures provide an overview of redesigns based on static network representations. The journey-level measures are based on journey trajectories generated with a customized routing algorithm, assessing the distributive effects of the redesign. The framework is applied to public transport networks from nine cities with recently implemented redesigns, namely Amsterdam, Auckland, Austin, Baltimore, Columbus, Helsinki, Houston, Indianapolis, and Wellington. Results indicate that the redesigns have improved navigability both from a system-level and user perspective in general. However, in some cases, improvements in navigability come at the cost of increased travel time and number of transfers. Furthermore, the results suggest that the redesigns have differing emphasis within the regions, for different times of day, and for different aspects of network structure. The results are discussed both from the perspective of the case findings and for drawing more general planning and policy recommendations. Finally, this research provides a basis for further transdisciplinary approaches, encouraging connections between transport modeling and complex networks approaches.

Towards a process-driven network analysis

A popular approach for understanding complex systems is a network analytic one: the system’s entities and their interactions are represented by a graph structure such that readily available methods suitable for graph structures can be applied. A network representation of a system enables the analysis of indirect effects: if A has an impact on B, and B has an impact on C, then, A also has an impact on C. This is often due to some kind of process flowing through the network, for example, pieces of informations or viral infections in social systems, passenger flows in transportation systems, or traded goods in economic systems. We argue that taking into account the actual usage of the system additionally to the static network representation of the system can yield interesting insights: first, the network representation and applicable network methods cannot be chosen independently from the network process of interest (Borgatti 2005; Dorn et al. 2012; Zweig 2016; Butts 2009). Therefore, focussing on the relevant network process in an early stage of the research project helps to determine suitable network representations and methods in order to obtain meaningful results (we call this approach process-driven network analysis). Second, many network methods assume that the spreading of some entity follows shortest or random paths. However, we show that not all flows are well approximated by this. In these cases, incorporating the network usage creates a real addition of knowledge to the static aggregated network representation.NoteThis is an extended and revised version of a conference article (Bockholt and Zweig 2019), published and presented at COMPLEX NETWORKS 2019.

A new irregular cellular learning automata-based evolutionary computation for time series link prediction in social networks

Link prediction (LP), as an attempt to predict event-based future connections within a network, is the main task of social network analysis (SNA). Accordingly, common LP approaches to forecast future connections utilize similarity metrics of non-connected links in a static network representation. A general shortcoming of most existing research studies in this field is that they tap the present condition of a system and fail to take any temporal events into account. Moreover; social networks are innately evolutionary since they are assumed to be online, non-deterministic, and unforeseeable in most applications. Consequently, it is not appropriate to employ deterministic models for examining actual social network problems. With regard to time-series LP (TSLP) problems, temporal evolution of connection incidence is correspondingly exploited to predict connection chances at a particular time. In this paper, a new TSLP method based on irregular cellular learning automaton (ICLA) and evolutionary computation (EC) is proposed. In the evolutionary procedure suggested here, each vertex (i.e. cell) includes a genome as well as a set of learning automata (LAs). Accordingly, the genome residing in a cell represents predicted links for the corresponding cell. Local information among cells in successive time 1 to T in the network is then analyzed to predict future connections in time T + 1. According to the distributed feature of the recommended approach, each genome is locally developed by a local search. The experiments in this study via e-mail and co-authorship networks ultimately show that the proposed algorithm leads to remarkable outcomes in predicting future connections.

Succinct Representation of Dynamic Networks

Many network analysis tasks like classification over nodes require careful efforts in engineering features used by learning algorithms. Most of recent studies have been made and succeeded in the field of static network representation learning. However, real-world networks are often dynamic and little work has been done on how to describe dynamic networks. In this work, we pose the problem of condensing dynamic networks and introduce SuRep, an encoding-decoding framework which utilizes matrix factorization technique to derive a succinct representation of a dynamic network in any stationary phase. We show that the succinct representation method can uncover the invariant structural properties in the network evolution and derive dense feature representations of the nodes as the byproduct. This method can be easily extended to dynamic attribute networks. For experiments on detecting change points in dynamic networks and network classification with real-world datasets we demonstrate SuRep's potential for capturing latent patterns among nodes.

Alignment of dynamic networks.

MotivationNetwork alignment (NA) aims to find a node mapping that conserves similar regions between compared networks. NA is applicable to many fields, including computational biology, where NA can guide the transfer of biological knowledge from well- to poorly-studied species across aligned network regions. Existing NA methods can only align static networks. However, most complex real-world systems evolve over time and should thus be modeled as dynamic networks. We hypothesize that aligning dynamic network representations of evolving systems will produce superior alignments compared to aligning the systems’ static network representations, as is currently done.ResultsFor this purpose, we introduce the first ever dynamic NA method, DynaMAGNA ++. This proof-of-concept dynamic NA method is an extension of a state-of-the-art static NA method, MAGNA++. Even though both MAGNA++ and DynaMAGNA++ optimize edge as well as node conservation across the aligned networks, MAGNA++ conserves static edges and similarity between static node neighborhoods, while DynaMAGNA++ conserves dynamic edges (events) and similarity between evolving node neighborhoods. For this purpose, we introduce the first ever measure of dynamic edge conservation and rely on our recent measure of dynamic node conservation. Importantly, the two dynamic conservation measures can be optimized with any state-of-the-art NA method and not just MAGNA++. We confirm our hypothesis that dynamic NA is superior to static NA, on synthetic and real-world networks, in computational biology and social domains. DynaMAGNA++ is parallelized and has a user-friendly graphical interface.Availability and implementation http://nd.edu/∼cone/DynaMAGNA++/.Supplementary information Supplementary data are available at Bioinformatics online.

Computation and analysis of temporal betweenness in a knowledge mobilization network

BackgroundHighly dynamic social networks, where connectivity continuously changes in time, are becoming more and more pervasive. Knowledge mobilization, which refers to the use of knowledge toward the achievement of goals, is one of the many examples of dynamic social networks. Despite the wide use and extensive study of dynamic networks, their temporal component is often neglected in social network analysis, and statistical measures are usually performed on static network representations. As a result, measures of importance (like betweenness centrality) typically do not reveal the temporal role of the entities involved. Our goal is to contribute to fill this limitation by proposing a form of temporal betweenness measure (foremost betweenness).MethodsOur method is analytical as well as experimental: we design an algorithm to compute foremost betweenness, and we apply it to a case study to analyze a knowledge mobilization network.ResultsWe propose a form of temporal betweenness measure (foremost betweenness) to analyze a knowledge mobilization network and we introduce, for the first time, an algorithm to compute exact foremost betweenness. We then show that this measure, which explicitly takes time into account, allows us to detect centrality roles that were completely hidden in the classical statistical analysis. In particular, we uncover nodes whose static centrality was negligible, but whose temporal role might instead be important to accelerate mobilization flow in the network. We also observe the reverse behavior by detecting nodes with high static centrality, whose role as temporal bridges is instead very low.ConclusionIn this paper, we focus on a form of temporal betweenness designed to detect accelerators in dynamic networks. By revealing potentially important temporal roles, this study is a first step toward a better understanding of the impact of time in social networks and opens the road to further investigation.

NetworkViewer: visualizing biochemical reaction networks with embedded rendering of molecular interaction rules.

BackgroundNetwork representations of cell-biological signaling processes frequently contain large numbers of interacting molecular and multi-molecular components that can exist in, and switch between, multiple biochemical and/or structural states. In addition, the interaction categories (associations, dissociations and transformations) in such networks cannot satisfactorily be mapped onto simple arrows connecting pairs of components since their specifications involve information such as reaction rates and conditions with regard to the states of the interacting components. This leads to the challenge of having to reconcile competing objectives: providing a high-level overview without omitting relevant information, and showing interaction specifics while not overwhelming users with too much detail displayed simultaneously. This problem is typically addressed by splitting the information required to understand a reaction network model into several categories that are rendered separately through combinations of visualizations and/or textual and tabular elements, requiring modelers to consult several sources to obtain comprehensive insights into the underlying assumptions of the model.ResultsWe report the development of an application, the Simmune NetworkViewer, that visualizes biochemical reaction networks using iconographic representations of protein interactions and the conditions under which the interactions take place using the same symbols that were used to specify the underlying model with the Simmune Modeler. This approach not only provides a coherent model representation but, moreover, following the principle of “overview first, zoom and filter, then details-on-demand,” can generate an overview visualization of the global network and, upon user request, presents more detailed views of local sub-networks and the underlying reaction rules for selected interactions. This visual integration of information would be difficult to achieve with static network representations or approaches that use scripted model specifications without offering simple but detailed symbolic representations of molecular interactions, their conditions and consequences in terms of biochemical modifications.ConclusionsThe Simmune NetworkViewer provides concise, yet comprehensive visualizations of reaction networks created in the Simmune framework. In the near future, by adopting the upcoming SBML standard for encoding multi-component, multi-state molecular complexes and their interactions as input, the NetworkViewer will, moreover, be able to offer such visualization for any rule-based model that can be exported to that standard.

Introducing exponential random graph models for visibility networks

Archaeological network analysts often represent archaeological data as static networks and explore their structure. However, most networks changed through time and static network representations do not allow archaeologists to test assumptions about the dynamic processes driving this change. The study of visibility networks in archaeology is a good example of this. Archaeologists propose hypotheses of the role of lines of sight between settlements, which imply dynamic processes for the establishment of the observed visibility networks. However, commonly used methods do not allow us to evaluate these hypotheses. In this paper we introduce exponential random graph modelling (ERGM) as a method for bridging static and dynamic approaches to interpreting visibility networks. This method offers a number of advantages: (1) it explicitly addresses the assumptions inherent in visibility network creation about what relationships between nodes mean and the types of processes they allow for; (2) it allows one to investigate the range of network structures that these assumptions give rise to; and (3) it explores the dynamic processes that might have led to observed networks. This method is used to evaluate hypotheses of the role of lines of sight in facilitating visual control and communication during the later Iron Age in Southern Spain. This study shows that ERGMs can be used as a reflective technique to evaluate competing hypotheses, and that ERGM results subsequently require more contextualised evaluation. Future work on ERGMs should focus on incorporating geographical constraints to further enhance its potential for studying visibility networks.

Contagion dynamics in time-varying metapopulation networks

The metapopulation framework is adopted in a wide array of disciplines to describe systems of well separated yet connected subpopulations. The subgroups or patches are often represented as nodes in a network whose links represent the migration routes among them. The connections have been so far mostly considered as static, but in general evolve in time. Here we address this case by investigating simple contagion processes on time-varying metapopulation networks. We focus on the SIR process and determine analytically the mobility threshold for the onset of an epidemic spreading in the framework of activity-driven network models. We find profound differences from the case of static networks. The threshold is entirely described by the dynamical parameters defining the average number of instantaneously migrating individuals and does not depend on the properties of the static network representation. Remarkably, the diffusion and contagion processes are slower in time-varying graphs than in their aggregated static counterparts, the mobility threshold being even two orders of magnitude larger in the first case. The presented results confirm the importance of considering the time-varying nature of complex networks.

Representing the UK's cattle herd as static and dynamic networks

Network models are increasingly being used to understand the spread of diseases through sparsely connected populations, with particular interest in the impact of animal movements upon the dynamics of infectious diseases. Detailed data collected by the UK government on the movement of cattle may be represented as a network, where animal holdings are nodes, and an edge is drawn between nodes where a movement of animals has occurred. These network representations may vary from a simple static representation, to a more complex, fully dynamic one where daily movements are explicitly captured. Using stochastic disease simulations, a wide range of network representations of the UK cattle herd are compared. We find that the simpler static network representations are often deficient when compared with a fully dynamic representation, and should therefore be used only with caution in epidemiological modelling. In particular, due to temporal structures within the dynamic network, static networks consistently fail to capture the predicted epidemic behaviour associated with dynamic networks even when parameterized to match early growth rates.

Disease dynamics over very different time-scales: foot-and-mouth disease and scrapie on the network of livestock movements in the UK.

We analyse the relationship between the network of livestock movements in the UK and the dynamics of two diseases: foot-and-mouth disease (FMD), which has an incubation period of days, and scrapie, which incubates over years. For FMD, the time-scale of expected epidemics is similar to the time-scale of the evolution of the network. We argue that, under appropriate conditions, a static network analysis can be an appropriate tool for gaining insights into disease dynamics even when the relevant time-scales are similar, as with FMD. We show that a subclass of ‘linkage moves’ maintains the network structure, and so removing these links has a dramatic effect on the number of potentially infected farms, an effect corroborated by simulations. In contrast, because scrapie has a low probability of transmission per contact and a long incubation period, a static network representation is probably appropriate; however, the signature of the network in the pattern of transmission is likely to be faint. Scrapie-notifying farms were more likely to be associated with each other via trading at markets than were control farms; however, network community structure proves to be less representative of prevalence patterns than geographical region. These contradictory indicators emphasize that appropriate observation time frames and good discrimination among types of potentially infectious contacts are vital in order for network analysis to be a valuable epidemiological tool.