Skip to contents

Build One or Multiple Networks from Bipartite Links

Source: R/build_dynamic_networks.R
build_dynamic_networks.Rd

[Experimental]

build_dynamic_networks() builds one or several tbl_graph networks from a node table (source_id) and a bipartite link table (source_id -> target_id). build_network() is a wrapper for a single network.

It supports two backbone extraction methods:

  • structured filtering using coupling/cooccurrence measures from biblionetwork;

  • statistical filtering using null models from backbone (Neal 2022) .

The function can build a single network or multiple networks across time windows.

Usage

build_dynamic_networks(
  nodes,
  directed_edges,
  source_id,
  target_id,
  time_variable = NULL,
  time_window = NULL,
  projection_method = c("structured", "statistical"),
  model = c("sdsm", "fdsm", "fixedfill", "fixedrow", "fixedcol"),
  alpha = NULL,
  cooccurrence_method = c("coupling_angle", "coupling_strength", "coupling_similarity"),
  edges_threshold = 1,
  overlapping_window = FALSE,
  compute_size = FALSE,
  keep_singleton = FALSE,
  filter_components = FALSE,
  ...,
  backbone_args = list(),
  verbose = TRUE
)

Arguments

nodes

Table of nodes and their metadata. One row per node. For example, a table of articles with identifiers, authors, publication year, etc.

directed_edges

Table of bipartite links between source_id nodes and target_id entities (e.g., article -> reference, author -> paper).

source_id

Quoted name of the source-side node identifier.

target_id

Quoted name of the target-side identifier linked to each source node.

time_variable

Optional name of the column with a temporal variable (e.g., publication year).

time_window

Optional size of the time window (in units of time_variable) to construct temporal networks.

projection_method

method used to extract the network backbone. Choose between:

  • "structured": uses cooccurrence measures from the biblionetwork package;

  • "statistical": uses statistical models from the backbone package. Defaults to "structured". The "statistical" method can be computationally slow on large networks.

model

Statistical null model from backbone: one of "sdsm", "fdsm", "fixedfill", "fixedrow", "fixedcol". Required if projection_method = "statistical".

alpha

Significance threshold for statistical backbone filtering. Required if projection_method = "statistical". Lower values keep fewer edges.

cooccurrence_method

For projection_method = "structured", choose the coupling method:

  • "coupling_angle"

  • "coupling_strength";

  • "coupling_similarity".

edges_threshold

Threshold used to filter weak edges in structured mode.

overlapping_window

Logical. If TRUE, builds networks using rolling time windows.

compute_size

Logical. If TRUE, computes the number of incoming edges per node (e.g., citation count).

keep_singleton

Logical. If FALSE, removes nodes with no edges in the final network.

filter_components

Logical. If TRUE, keeps only the main component(s) using networkflow::filter_components().

...

Additional arguments passed to filter_components().

backbone_args

Optional list of additional arguments passed to the backbone extraction call. If backbone_args includes alpha or model, those values override function arguments.

verbose

Logical. If TRUE, displays progress messages.

Value

  • A single tidygraph object if time_window is NULL;

  • A list of tidygraph objects (one per time window) otherwise.

Details

The function uses bipartite links (source_id -> target_id) to produce source-side networks.

If time_variable and time_window are provided, it builds one network per time window (rolling or non-overlapping). Otherwise it builds a single network.

projection_method = "structured" applies coupling/cooccurrence filtering. projection_method = "statistical" applies a statistical backbone model.

References

Neal ZP (2022). “backbone: An R Package to Extract Network Backbones.” PLOS ONE, 17, e0269137. doi:10.1371/journal.pone.0269137 .

See also

biblionetwork::biblio_coupling(), backbone::backbone()

Examples

library(networkflow)

nodes <- networkflow::Nodes_stagflation |>
  dplyr::filter(source_type == "Stagflation")

references <- networkflow::Ref_stagflation

# Structured backbone (cooccurrence)
net_structured <- build_dynamic_networks(
nodes = nodes,
directed_edges = references,
source_id = "source_id",
target_id = "target_id",
time_variable = "source_year",
time_window = 20,
projection_method = "structured",
cooccurrence_method = "coupling_similarity",
edges_threshold = 1
)
#> ℹ Backbone method selected: structured
#> ℹ Keep_singleton == FALSE: removing the nodes that are alone with no edge. 
#> Warning: Your last network is shorter than the other(s) because the cutting by time
#> window does not give a round count. The last time unity in your data is 2013,
#> but the upper limit of your last time window is 2014.
#> 
#> ── Generation of the network for the 1975-1994 time window. ────────────────────
#> 
#> ── Generation of the network for the 1995-2014 time window. ────────────────────

# Statistical backbone (backbone package)
net_statistical <- build_dynamic_networks(
nodes = nodes,
directed_edges = references,
source_id = "source_id",
target_id = "target_id",
time_variable = "source_year",
time_window = 20,
projection_method = "statistical",
model = "sdsm",
alpha = 0.05,
backbone_args = list(mtc = "holm")
)
#> ℹ Backbone method selected: statistical
#> ℹ Keep_singleton == FALSE: removing the nodes that are alone with no edge. 
#> Warning: Your last network is shorter than the other(s) because the cutting by time
#> window does not give a round count. The last time unity in your data is 2013,
#> but the upper limit of your last time window is 2014.
#> 
#> ── Generation of the network for the 1975-1994 time window. ────────────────────
#> 
#> ── Generation of the network for the 1995-2014 time window. ────────────────────