 Research
 Open Access
 Published:
Mapping change in higherorder networks with multilevel and overlapping communities
Applied Network Science volume 8, Article number: 42 (2023)
Abstract
New network models of complex systems use layers, state nodes, or hyperedges to capture higherorder interactions and dynamics. Simplifying how the higherorder networks change over time or depending on the network model would be easy with alluvial diagrams, which visualize community splits and merges between networks. However, alluvial diagrams were developed for networks with regular nodes assigned to nonoverlapping flat communities. How should they be defined for nodes in layers, state nodes, or hyperedges? How can they depict multilevel, overlapping communities? Here we generalize alluvial diagrams to map change in higherorder networks and provide an interactive tool for anyone to generate alluvial diagrams. We use the alluvial diagram generator in three case studies to illustrate significant changes in the organization of science, the effect of modeling network flows with memory in a citation network and distinguishing multidisciplinary from fieldspecific journals, and the effects of multilayer representation of a collaboration hypergraph.
Introduction
Complex systems are inherently dynamic. Their components influence each other through various informational and physical processes, changing interaction patterns over time. Researchers represent these interactions with networks (Edler et al. 2017; Calatayud et al. 2020; Farage et al. 2021; Calatayud et al. 2021; Neuman 2022; Edler et al. 2022a; Rojas et al. 2022) and simplify their organization with communitydetection algorithms (Rosvall and Bergstrom 2008; Fortunato 2010; Schaub et al. 2017; Traag et al. 2019; Peixoto 2019). For example, communitydetection algorithms that model the various processes as flows on networks assign nodes to possibly nested modules of typically densely connected nodes, among which the network flows persist relatively long (Rosvall and Bergstrom 2008). Identifying modules in multiple networks with shared nodes enables exploring organizational changes when the systems they represent change over time or between states: Modules merge and split when groups in students’ social networks form and dissolve during school days, or new research fields emerge when old fields fuse or break and move apart. Various summary statistics can quantify these structural changes (Danon et al. 2005; Amelio and Pizzuti 2017; Newman et al. 2020), but they destroy essential information about how the networks change.
Alluvial diagrams with modules represented as stacks of blocks joined by stream fields were introduced to reveal network organizational changes by depicting merging and splitting modules (Rosvall and Bergstrom 2010). Researchers have successfully used them to map shifting regional tendencies in urban networks (Liu et al. 2013), study dynamics of hot topics in research fields (Ruan et al. 2017; Pal et al. 2022), track changing bitcoin user activity (Cazabet et al. 2017), and explore evolving media channel preferences across crisis phases (Petrun Sayers et al. 2021). Generating alluvial diagrams requires dedicated software to remove tedious manual work. However, current applications to generate alluvial diagrams work only for standard networks partitioned into modules.
Today researchers use temporal, multilayer, and memory networks to capture interactions in complex systems with higher accuracy (Kivelä et al. 2014; Rosvall et al. 2014; De Domenico et al. 2015, 2016; Xu et al. 2016; Lambiotte et al. 2019) and multilevel modular solutions to reveal more regularities in their organization (Rosvall and Bergstrom 2011; Peixoto 2014). Multilayer networks can represent networks over time with links in timewindowed layers. Memory networks can represent higherorder network flow models where the transition rates depend on the current node and previously visited nodes. Both representations enable overlapping modules. Mapping change in these rich network representations requires generalizing alluvial diagrams and their generators to higherorder networks with multilevel and overlapping modular solutions.
Here we introduce alluvial diagrams for multilayer and memory networks with multilevel and overlapping modular solutions. We demonstrate a new alluvial generator for higherorder networks available for anyone to use at https://www.mapequation.org/alluvial (Holmgren et al. 2022a), and illustrate how we use it in three case studies revealing: significant changes in the multilevel organization of science over six years using parametric bootstrap resampling, multidisciplinary journals in a secondorder network representation of citation flows, and the effects of multilayer representation of a collaboration hypergraph.
Methods
Alluvial diagrams depict changes in the modular composition between networks with stacks of blocks representing the modules (Fig. 1). Each block’s height is proportional to the flow volume of the corresponding module—the total visit probability of all nodes in the module. To highlight structural change between multiple networks, a vertical stack of blocks represent each network’s modular structure, and horizontal stream fields connect blocks that share nodes across neighboring networks. Like block heights, streamfield heights are proportional to the flow volume of the node overlap between corresponding modules. To reduce clutter, we order stream fields to minimize their overlap.
We use Infomap to search for multilevel modular structures with nested submodules (Edler et al. 2022b; Rosvall and Bergstrom 2011). Throughout the paper, we use multilevel to denote partitions with nested submodules as illustrated in Fig. 1d–e, and multilayer to denote the network type that stratifies connections between nodes into different layers. Infomap optimizes the map equation, the average perstep codelength on a modular description of a random walk modeling network flows (Rosvall and Bergstrom 2008). The modules are groups of nodes where the random walker spends a relatively long time compared to exiting it and entering other modules. While we focus on modules derived from Infomap, alluvial diagrams work with output from any community detection or hierarchical data clustering method.
Mapping change in networks with multilevel communities
We extend alluvial diagrams to multilevel network partitions by nesting submodules in supermodules with adaptive module distances. The right multilayer stack of the schematic alluvial diagram in Fig. 1c illustrates. In the spirit of cartography, we put blocks corresponding to the toplevel modules in a bottom layer to highlight the largescale organization and provide a cleaner visualization. Optionally, we display finerlevel structures in layers above the bottom layer. The right multilayer stack in Fig. 1c expands the left stack’s single layer with one such extra layer corresponding to the four submodules of the multilevel modular solution. To show that deeper submodules are more closely related than their larger parent modules, we draw sibling submodules closer together than other modules. Specifically, we halve the distance between two adjacent modules for each level down in the multilevel solution.
Multilevel significance clustering
To separate trends from mere noise in the module assignments, we extend the significance clustering method described in ref. Rosvall and Bergstrom (2010) to multilevel partitions. The approach has three main steps: First, we search for optimal multilevel partitions for each network using Infomap. Then, to assess these partitions’ robustness to slight perturbations in the data, we create a large number of independent bootstrap networks. For each bootstrap network, we search for the optimal multilevel partition using Infomap as for the original network. Finally, we summarize the variability in the bootstrap partitions by applying the significance clustering method introduced in ref. Rosvall and Bergstrom (2010) extended to multilevel partitions. For each level in the multilevel solution of the original network, we search for the largest subset of nodes in each module or submodule that are also clustered together in at least a fraction p of solutions obtained from the parametric bootstrap procedure.
Searching for significant subsets in multilevel solutions is computationally more demanding than for ordinary twolevel partitions. To improve the performance, we trivially parallelize the algorithm by running each module or submodule in separate threads.
Mapping change in higherorder networks
We generalize alluvial diagrams to multilayer and memory networks. Multilayer networks can model different modes of interaction or interactions that change over time in different layers. Memory networks can model dynamics that depend on from where the flows come. Infomap represents both higherorder networks with socalled state nodes (Edler and Bohlin 2017). In higherorder networks, we call ordinary nodes physical nodes to distinguish them from state nodes. In a multilayer network, one state node for each physical node and layer represents the physical node in the layer (De Domenico et al. 2015). In a secondorder memory network with memory of the previous step, one state node for each physical node and incoming link represents the physical node for flows incoming along that link (Rosvall et al. 2014) In this way, the order of a memory network corresponds to the order of a Markov process, the first order being regular memoryless Markov dynamics. Physical nodes with multiple state nodes and different outgoing links can model higherorder dynamics on the network.
In theory, using alluvial diagrams for higherorder networks is no different than for ordinary networks. In practice, the many possible combinations of first and higherorder networks, memory networks with different memory, and multilayer networks with different layers make it challenging to determine node equality in different networks because we need to match nodes across networks to draw stream fields between modules. While alluvial diagrams require networks to share a significant fraction of physical nodes, we also need their state nodes to match since they are the smallest components of higherorder networks. With no universal solution to this nodematching problem, we discuss some challenges and how we choose to solve them.
First and higherorder networks
Alluvial diagrams with first and higherorder networks require matching different node types: Firstorder networks have only physical nodes, but higherorder networks have physical nodes and state nodes. We illustrate this schematically in Fig. 2 with a firstorder network in Fig. 2a and a higherorder network with state nodes as smaller circles inside the physical nodes in Fig. 2c. We consider only hard module boundaries in the firstorder network, whereas modules overlap in the higherorder network when physical nodes’ state nodes are assigned to different modules. In Fig. 2c, the modules overlap in the physical nodes containing the purple and blue state nodes.
As we need a onetoone match across networks to draw stream fields, we cannot match all state nodes in the higherorder network to one firstorder node. To overcome this problem, we first split the firstorder nodes into pseudostate nodes, which we depict with small dashed circles in Fig. 2a. We create as many pseudostate nodes as there are state nodes in the matching physical node in the higherorder network. Then, we divide firstorder node i’s flow volume \(\pi _i^{(1)}\) among its pseudostates \(\alpha\) proportionally to their matching state nodes’ fraction of the flow \(\pi _{i\alpha }^{(2)}\) as
This procedure gives a onetoone match between nodes in first and higherorder networks, and we can draw multiple stream fields from a single firstorder node (Fig. 2b).
Memory networks
Drawing alluvial diagrams for memory networks requires matching state nodes representing corresponding memory in different networks. We match state nodes across networks by encoding their memory in their ids such that state nodes representing the same memory share the same id in different networks. As long as the networks are not too large, we can encode memory of order n into a single binary number by dividing the binary number into n parts: We divide the number into two parts in a secondorder memory network with memory of the previous step. With N physical nodes, we use the \(b = \lceil \log _2 N \rceil\) most significant bits of the state id to encode the previously visited node i and the b least significant bits to encode the currently visited node j, resulting in the state id
where \(\ll\) is the arithmetic leftshift operator and \(\vee\) is the logical or. For example, we encode the link from physical node 2 to physical node 3 along the path represented by the trigram \(1\rightarrow 2\rightarrow 3\) as
resulting in the directed link \(10 \rightarrow 19\) between state nodes 10 and 19. This encoding scheme works for up to \(N = 2^{16} = 65,536\) physical nodes with 32bit ids and secondorder memory.
Multilayer networks
When comparing multiple multilayer networks with N layers, we encode the physical node i in layer l with id
where N is the largest layer id represented with \(b = \lceil \log _2 N \rceil\) bits. For multilayer networks, this encoding scheme is available in Infomap using the flag matchablemultilayerids N.
Alluvial diagrams can also visualize the layers of multilayer networks, each as a separate network. In this case, node matching is trivial as physical nodes are unique in each layer. The stream fields then connect modules that span layers.
Alluvial diagram generator
We have implemented an interactive web application that generates alluvial diagrams, available for anyone to use at https://www.mapequation.org/alluvial. We implemented it as a clientside web application to enable researchers to use our application without programming experience or those working with sensitive data. All code runs locally in the user’s web browser, and the web application does not store or upload network data to any server. We implemented it using TypeScript and React, and we display the diagrams using scalable vector graphics (SVG) (see Additional file 1: Fig. S2 in the SI for how we model the data structures).
While the most efficient community detection pipeline is to run the standalone C++ version of Infomap and load the resulting partitions, we have embedded a version of Infomap compiled to JavaScript with Emscripten (Zakai 2011). This embedded Infomap version supports the same network inputs as C++ Infomap, but only a subset of Infomap’s features, including reading directed or undirected input, choosing the number of optimization trials, and searching for multilevel or twolevel solutions (Fig. 3). We defer the specification of input formats to Additional file 1: section SI.1. We also support loading solutions from Infomap Online (Holmgren et al. 2022b), a fully featured webbased version of Infomap.
With loaded networks, the interface shows the user a toplevel view of the alluvial diagram (Additional file 1: Fig. S1). The user can manipulate the diagram in several ways: expand modules to reveal their submodules, reorganize networks and modules for clarity, highlight modules or individual nodes with different colors, and change the diagram width and height. While we have implemented the features and use cases we think most researchers use, we can imagine feature requests for specific use cases. By supporting export to SVG, researchers can modify the diagrams to their needs in any vector graphics application.
Results
We highlight different visualization challenges in three case studies using multilevel, higherorder, and multilayer networks. In all cases, we use Infomap to identify optimal multilevel solutions using unrecorded teleportation to links with minimal impact from the teleportation rate on the results (Lambiotte and Rosvall 2012).
Robust multilevel citation networks
First, we highlight the multilevel organization of science into research areas and fields. We use data from ThomsonReuters Journal Citation Reports (Rosvall and Bergstrom 2010). The data include citations between journals published from 2001–2007, divided into four twoyear periods. The networks have, on average, 7, 490 nodes representing journals and 586, 295 integerweighted links representing the citation flow between them. For each year, we use Infomap with 100 optimization trials to search for the optimal multilevel solution. We use the multilevel significance clustering approach described in the “Methods” section to assess the solution’s robustness to slight perturbations in the data. First, we create 1000 independent bootstrap networks by sampling each citation weight \(w_{uv}\) from a Poisson distribution, \({\hat{w}}_{uv} \sim \text {Poisson}(w_{uv})\). Then, we use Infomap to search for the optimal multilevel solution for each bootstrap network. The bootstrap solutions have similar codelengths, with a variance of around \(10^{5}\). Finally, we use the significance clustering algorithm to search for the largest fraction of nodes clustered together in at least a fraction \(p=0.95\) of the bootstrap solutions.
The resulting multilevel partitions organize science into research areas, further divided into research fields (Fig. 4). With the multilevel solution and unrecorded teleportation scheme, we do not exactly reproduce the results presented in Ref. Rosvall and Bergstrom (2010). The life sciences show higher diversification, with more significant research fields and lower citation flows in molecular and cell biology containing J. Biol. Chem., Nature, PNAS, Science, Cell, and so on.
First and secondorder citation networks
In the second case study, we visualize the effects of using higherorder network models with alluvial diagrams. We organize the citation data from the ThomsonReuters Journal Citation Reports into citation pathways (Persson et al. 2016; Wang and Waltman 2016). The data contain citations between articles published from 2007 to 2012 in the \(10\,000\) journals with the highest impact factor, and all citation pathways contain at least one article published in 2009. When aggregated to journals, we are left with 69, 738, 205 weighted trigrams.
To study the effect of a secondorder model, we model the data using both first and secondorder Markov chains. We create a firstorder network by discarding the first step from each trigram. For example, the trigram \(i\rightarrow j\rightarrow k\) with weight w becomes the directed link \(j\rightarrow k\) with the same weight, resulting in 69 million links between the 10, 000 nodes. Using the complete trigram data, we create a secondorder network. For each trigram \(i\rightarrow j\rightarrow k\) with weight w, we create two state nodes if they do not already exist:

\(\alpha _{i \rightarrow j}\) in physical node j representing the memory of coming from i,

\(\alpha _{j \rightarrow k}\) in physical node k representing the memory of coming from j.
We connect the state nodes with a directed link \(\alpha _{i \rightarrow j} \rightarrow \alpha _{j \rightarrow k}\) with weight w. The resulting secondorder network has around 3.9 million state nodes connected by 69 million links.
Because the secondorder network has two orders of magnitude more state nodes than the firstorder network has physical nodes, the community detection search space is much larger, significantly impacting the computational time. The firstorder network takes around two minutes for ten optimization trials, while the secondorder network takes around nine hours for the same task on a 2021 MacBook Pro with the M1 Max CPU and 32 GB of RAM. The resulting firstorder partition has codelength \(L^{(1)} = 8.44\) bits, five top modules, and four levels. The secondorder partition has codelength \(L^{(2)} = 7.83\) bits, around 4, 700 top modules, and five levels. Although the secondorder partition has many top modules, most are tiny, containing only one or a few state nodes. To downplay small modules at the fringe of the citation data, we compare the partition’s effective number of top modules using the perplexity \(M_{\text {eff}} = 2^{H(M)}\), with Shannon entropy
where \(\pi _m = \sum _{i \in m} \pi _i\) is the total flow volume of the nodes i in module m. With this metric, the first and secondorder partitions are similar with \(M_\text {eff}^{(1)} = 2.35\) and \(M_\text {eff}^{(2)} = 2.73\) effective top modules, respectively.
After detecting communities, we aggregate redundant state nodes in the secondorder network before visualization for better performance. We lump state nodes in the same physical node and leaf module and aggregate their flows, reducing the number of states to visualize from 3.9 million to 355 thousand. After lumping, we remove any state nodes with zero flow that would not contribute to the alluvial diagram layout, further reducing the number of states to 271 thousand. Then, we create pseudostates in the firstorder network to match the higherorder state nodes. After this step, both networks contain 271 thousand state nodes. In the firstorder network, all state nodes are in the same module as their physical node.
The alluvial diagram shows how the secondorder model separates cosmology and astrophysics—journals clustered together with Astrophys J. and Phys. Rev. D. – from the physical sciences (Fig. 5). The cell and molecular biology submodule containing Nature, PNAS, and Science grows, and the multidisciplinary journals’ submodules in the life sciences divide into smaller modules.
Above all, Nature, Science, and PNAS are all recognized as multidisciplinary journals represented in multiple research fields. To quantify how a higherorder model captures their citation flows, we investigate in how many research fields journals are present. Since a single research field dominates most journals’ citation flows, we measure the effective number of research fields. With journal i’s moduleaggregated state node flow \(\pmb {\pi }_i = \{ \pi _{i\alpha } \}\), we calculate its effective number of research fields \(r_i = 2^{H(\pmb {\pi }_i)}\) with the entropy \(H(\pmb {\pi }_i) =  \sum _\alpha \pi _{i\alpha } \log _2 \pi _{i\alpha }\). With this metric, the most overlapping journals are Bratislava Medical J., Quality and Quantity, and Harvard Business Review – tiny journals with only around \(10^{4}\) percent of the total citation flow. To highlight prominent, multidisciplinary journals and mesoscale changes in the citation flows, we weigh each journal’s effective number of research fields with its total citation flow \(\pi _i = \sum _\alpha \pi _{i\alpha }\) for a weighted overlap
The journals with the highest weighted overlap are Nature, Science, and PNAS (Fig. 6).
The life sciences contain more of the multidisciplinary citation flow than the other research areas. By aggregating the weighted overlap \(o_i\) on the leaf modules m,
around 60 percent of the 1000 most overlapping leaf modules are in the life sciences, followed by the physical sciences with 18 percent.
Collaboration hypergraph using different representations
Finally, we study how a hypergraph’s different firstorder and multilayer network representations affect the detected communities. We use a collaboration hypergraph extracted from the 734 references in the review article “Networks beyond pairwise interactions: structure and dynamics” (Eriksson et al. 2021; Battiston et al. 2020). The referenced articles form hyperedges linking their authors. These hyperedges overlap in those authors who authored multiple papers, with the largest connected component containing 361 author nodes V in 220 hyperedges E. We illustrate a small, schematic hypergraph in Fig. 7a, where the white circles represent authors and the larger, orange circles represent papers.
To model the flow of ideas among collaborators, we model a random walk on the hypergraph. Each hyperedge e has a weight \(\omega (e)\), and each node u has a hyperedgedependent weight \(\gamma _e(u)\). We denote u’s total incident hyperedge weight
and hyperedge e’s total node weight
A random walker moves from node u to v with these weights in three stages by Chitra and Raphael (2019): First, choosing hyperedge e among node u’s hyperedges E(u) with probability \(\frac{\omega (e)}{d(u)}\). Then, choosing one of the hyperedge e’s nodes v with probability \(\frac{\gamma _e(v)}{\delta (e)}\). And finally, moving to v.
For the collaboration hypergraph, we use article hyperedge weights \(w(e) = \ln (c + 1) + 1\) where c is the number of citations for that article in December 2020 (Eriksson et al. 2021). To model the author’s unequal contributions to articles, we use hyperedgedependent node weights (Chitra and Raphael 2019).
We weigh alphabetically sorted authors uniformly because their contributions are hard to determine.
From this hypergraph, we generate bipartite and multilayer hypergraph representations with identical node visit rates using the method described in Ref. Eriksson et al. (2021) (Fig. 7b–c). We represent walks on hypergraphs as a bipartite network by representing the hyperedges with hyperedge nodes, and the three stages become a twostep walk between the nodes at the bottom and the hyperedge nodes at the top in 7b. First, a step from a node u to a hyperedge node e,
and then a step from the hyperedge node to a node v,
To represent the random walk on a multilayer network, we project the threestage randomwalk process down to a onestep process on state nodes in separate layers. Each hyperedge e with weight \(\omega (e)\) forms a layer \(\alpha\) with weight \(\omega (\alpha )\). A state node \(u^{\alpha }\) represents u in each layer \(\alpha \in {E}(u)\) that contains the node (Fig. 7c). The transition rate between state node \(u^{\alpha }\) in layer \(\alpha\) and state node \(v^{\beta }\) in layer \(\beta\) is
With one state node per hyperedge layer that contains the node, the multilayer representation requires more nodes and links than the bipartite representation.
We also generate a multilayer network using a socalled hyperedgesimilarity model that increases the probability of a random walk staying among similar hyperedges (Eriksson et al. 2021). This model reinforces community structure with modules formed by similar sets of collaborators. We let Infomap search for optimal multilevel solutions in the three network representations. As before, we create pseudostate nodes in the bipartite network to match them with the multilayer networks’ state nodes.
The resulting partitions have effectively three or four levels. The toplevel organization is most coarsegrained for the bipartite representation and most finegrained for the hyperedgesimilarity representation (Fig. 8). Only the multilayer representation assigns the submodule “Peixoto” together with the top module in which Bianconi is the highestranking author. It also assigns Fortunato to a different top module than the hyperedgesimilarity partition. Finally, Bocaletti overlaps as the highestranking author in two submodules in the hyperedgesimilarity partition in the same top module as Bianconi.
Conclusions
We have extended alluvial diagrams to higherorder networks with multilevel and overlapping communities and implemented an interactive web application available for anyone to use. In three case studies, we have used alluvial diagrams to show how the multilevel organization of science changes over time, how a secondorder model compares to a firstorder model, and how different hypergraphflow equivalent networks influence the flow of ideas among network scientists.
We have focused on flowbased community detection using the map equation framework and the search algorithm Infomap. The generalized alluvial diagrams apply to any communitydetection algorithm and are particularly relevant for simplifying and highlighting complex multilevel and overlapping modular descriptions of large higherorder networks.
Availability of data and materials
The interactive web application is available at https://www.mapequation.org/alluvial, and its source code at https://github.com/mapequation/alluvialgenerator. The multilevel significance clustering code is available at https://github.com/mapequation/multilevelsignificanceclustering. All other relevant data and code are available at https://github.com/mapequation/mappingchange2.
References
Amelio A, Pizzuti C (2017) Correction for closeness: adjusting normalized mutual information measure for clustering comparison. Comput Intel 33:579
Battiston F, Cencetti G, Iacopini I, Latora V, Lucas M, Patania A, Young JG, Petri G (2020) Networks beyond pairwise interactions: structure and dynamics. Phys Rep 874:1
Calatayud J, Andivia E, Escudero A, Melián CJ, BernardoMadrid R, Stoffel M, Aponte C, Medina NG, MolinaVenegas R, Arnan X et al (2020) Positive associations among rare species and their persistence in ecological assemblages. Nat Ecol Evol 4:40
Calatayud J, Neuman M, Rojas A, Eriksson A, Rosvall M (2021) Regularities in species’ niches reveal the world’s climate regions. eLife 10
Cazabet R, Rym B, Matthieu L (2017) Tracking bitcoin users activity using community detection on a network of weak signals. In International conference on complex networks and their applications. Springer, pp 166–177
Chitra U, Raphael B (2019) Random walks on hypergraphs with edgedependent vertex weights. In: Proceedings of the 36th international conference on machine learning (PMLR), pp 1172–1181, iSSN: 26403498, https://proceedings.mlr.press/v97/chitra19a.html
Danon L, DiazGuilera A, Duch J, Arenas A (2005) Comparing community structure identification. J Stat Mech Theory Exp 2005:P09008
De Domenico M, Lancichinetti A, Arenas A, Rosvall M (2015) Identifying modular flows on multilayer networks reveals highly overlapping organization in interconnected systems. Phys Rev X 5:011027
De Domenico M, Granell C, Porter MA, Arenas A (2016) The physics of spreading processes in multilayer networks. Nat Phys 12:901
Edler D, Bohlin L et al (2017) Mapping higherorder network flows in memory and multilayer networks with Infomap. Algorithms 10:112
Edler D, Guedes T, Zizka A, Rosvall M, Antonelli A (2017) Infomap bioregions: interactive mapping of biogeographical regions from species distributions. Syst Biol 66:197
Edler D, Holmgren A, Rojas A, Rosvall M, Antonelli A (2022a) Infomap Bioregions 2: exploring the interplay between biogeography and evolution
Edler D, Holmgren A, Rosvall M (2022b) The MapEquation software package. https://mapequation.org
Eriksson A, Edler D, Rojas A, de Domenico M, Rosvall M (2021) How choosing randomwalk model and network representation matters for flowbased community detection in hypergraphs. Commun Phys 4:1
Farage C, Edler D, Eklöf A, Rosvall M, Pilosof S (2021) Identifying flow modules in ecological networks using Infomap. Methods Ecol Evol 12:778
Fortunato S (2010) Community detection in graphs. Phys Rep 486:75
Holmgren A, Edler D, Rosvall M (2022a) The MapEquation alluvial diagram generator. https://mapequation.org/alluvial
Holmgren A, Edler D, Rosvall M (2022b) Infomap online. https://mapequation.org/infomap
Kivelä M, Arenas A, Barthelemy M, Gleeson JP, Moreno Y, Porter MA (2014) Multilayer networks. J Compl Netw 2:203
Lambiotte R, Rosvall M (2012) Ranking and clustering of nodes in networks with smart teleportation. Phys Rev E 85:056107
Lambiotte R, Rosvall M, Scholtes I (2019) From networks to optimal higherorder models of complex systems. Nat Phys 15:313
Liu X, Derudder B, Csomós G, Taylor P (2013) Featured graphic. Mapping shifting hierarchical and regional tendencies in an urban network through alluvial diagrams. Environ Plann A 45:1005
Neuman M (2022) PISA data clusters reveal student and school inequality that affects results. Plos one 17:e0267040
Newman ME, Cantwell GT, Young JG (2020) Improved mutual information measure for clustering, classification, and community detection. Phys Rev E 101:042304
Pal R, Chopra H, Awasthi R, Bandhey H, Nagori A, Sethi T et al (2022) Predicting emerging themes in rapidly expanding COVID19 literature with unsupervised word embeddings and machine learning: evidencebased study. J Med Inter Res 24:e34067
Peixoto TP (2019) Bayesian stochastic blockmodeling. In: Advances in network clustering and blockmodeling pp. 289–332
Peixoto TP (2014) Hierarchical block structures and highresolution model selection in large networks. Phys Rev X 4:011047
Persson C, Bohlin L, Edler D, Rosvall M (2016) Maps of sparse Markov chains efficiently reveal community structure in network flows with memory. arXiv preprint arXiv:1606.08328
Petrun Sayers EL, Parker AM, Seelam R, Finucane ML (2021) How disasters drive media channel preferences: tracing news consumption before, during, and after Hurricane Harvey. J Contingen Crisis Manag 29:342
Rojas A, Eriksson A, Neuman M, Edler D, Blocker C, Rosvall M (2022) A natural history of networks: higherorder network modeling for paleobiology research. bioRxiv
Rosvall M, Bergstrom CT (2008) Maps of random walks on complex networks reveal community structure. Proc Natl Acad Sci 105:1118
Rosvall M, Bergstrom CT (2010) Mapping change in large networks. PloS One 5:e8694
Rosvall M, Bergstrom CT (2011) Multilevel compression of random walks on networks reveals hierarchical organization in large integrated systems. PloS One 6:e18209
Rosvall M, Esquivel AV, Lancichinetti A, West JD, Lambiotte R (2014) Memory in network flows and its effects on spreading dynamics and community detection. Nat Commun 5:1
Ruan W, Hou H, Hu Z (2017) Detecting dynamics of hot topics with alluvial diagrams: a timeline visualization. J Data Inf Sci 2:37
Schaub MT, Delvenne JC, Rosvall M, Lambiotte R (2017) The many facets of community detection in complex networks. Appl Netw Sci 2:1
Traag VA, Waltman L, Van Eck NJ (2019) From Louvain to Leiden: guaranteeing wellconnected communities. Sci Rep 9:1
Wang Q, Waltman L (2016) Largescale analysis of the accuracy of the journal classification systems of Web of Science and Scopus. J Inf 10:347
Xu J, Wickramarathne TL, Chawla NV (2016) Representing higherorder dependencies in networks. Sci Adv 2:e1600028
Zakai A (2011) Emscripten: an LLVMtoJavaScript compiler. In: Proceedings of the ACM international conference companion on object oriented programming systems languages and applications companion, pp 301–312
Acknowledgements
A.H. was supported by the Swedish Foundation for Strategic Research, Grant No. SB160089. D.E. and M.R. were supported by the Swedish Research Council (201600796).
Funding
Open access funding provided by Umea University.
Author information
Authors and Affiliations
Contributions
AH and MR devised the study. AH performed the experiments. AH and DE implemented the interactive web application. AH and MR wrote the manuscript. All authors edited and accepted the manuscript in its final form.
Corresponding author
Ethics declarations
Competing interests
The authors declare that they have no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Additional file 1.
Supplementary Information.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Holmgren, A., Edler, D. & Rosvall, M. Mapping change in higherorder networks with multilevel and overlapping communities. Appl Netw Sci 8, 42 (2023). https://doi.org/10.1007/s41109023005725
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s41109023005725