⚠️ This package is at a very early stage of development ⚠️
diskurs (German for discourse) is an R package to handle data on discourse networks in a very specific form.
The main goal of the package is to ensure validated data loading and basic transformations of discourse in graph structure as introduced in the sustainability.discourses project.
At the moment, it mainly exists to facilitate the reproduction of analyses conducted within the sustainability.discourses project. The eventual goal is to hopefully support anyone working with a similar data structure in the future.
diskurs is in most cases a thin wrapper or special use case around
tidygraph and
igraph.
You can install the development version of diskurs from the r-universe builds of diskurs with:
install.packages('diskurs', repos = c('https://urban-sustainability-lab-zurich.r-universe.dev', 'https://cloud.r-project.org'))
library(diskurs)
example_edgelist <- diskurs::edgelist_example
example_nodelist <- diskurs::nodelist_example
example_edgelist
#> from to stance timestamp
#> 1 1 4 support 2012-02-01
#> 2 1 4 support 2013-02-01
#> 3 1 5 irrelevant 2014-02-01
#> 4 1 4 support 2014-04-01
#> 5 2 5 opposition 2016-02-01
#> 6 3 4 support 2016-02-01
#> 7 3 5 support 2016-03-01
example_nodelist
#> nodeid name label mode
#> 1 1 actor1 Actor 1 actor
#> 2 2 actor2 Actor 2 actor
#> 3 3 actor3 Actress 3 actor
#> 4 4 statement1 The fascists will lose statement
#> 5 5 statement2 Owls are great statement
disc_g <- load_discourse_graph(edgelist = example_edgelist, nodelist = example_nodelist)
disc_g
#> ----------------------------------------
#> A discourse graph with 3 actors and 2
#> statements
#> ----------------------------------------
#> # A tbl_graph: 5 nodes and 7 edges
#> #
#> # A directed acyclic multigraph with 1 component
#> #
#> # Node Data: 5 × 4 (active)
#> nodeid name label mode
#> <int> <chr> <chr> <chr>
#> 1 1 actor1 Actor 1 actor
#> 2 2 actor2 Actor 2 actor
#> 3 3 actor3 Actress 3 actor
#> 4 4 statement1 The fascists will lose statement
#> 5 5 statement2 Owls are great statement
#> #
#> # Edge Data: 7 × 4
#> from to stance timestamp
#> <int> <int> <chr> <date>
#> 1 1 4 support 2012-02-01
#> 2 1 4 support 2013-02-01
#> 3 1 5 irrelevant 2014-02-01
#> # ℹ 4 more rows
Let’s look at it:
disc_g |> plot()
In many cases, working with the igraph or tidygraph object is advisable
disc_g |> get_igraph()
#> IGRAPH 327aa5f DN-- 5 7 --
#> + attr: nodeid (v/n), name (v/c), label (v/c), mode (v/c), stance
#> | (e/c), timestamp (e/n)
#> + edges from 327aa5f (vertex names):
#> [1] actor1->statement1 actor1->statement1 actor1->statement2 actor1->statement1
#> [5] actor2->statement2 actor3->statement1 actor3->statement2
This makes it possible to use the entire ecosystem provided by igraph:
disc_g |>
get_igraph() |>
plot()
Or tidygraph. You can extract tidygraph’s tbl_graph object and then use its verbs:
disc_g |>
get_tbl_graph() |>
tidygraph::activate(nodes) |>
dplyr::mutate(closeness = tidygraph::centrality_closeness())
#> # A tbl_graph: 5 nodes and 7 edges
#> #
#> # A directed acyclic multigraph with 1 component
#> #
#> # Node Data: 5 × 5 (active)
#> nodeid name label mode closeness
#> <int> <chr> <chr> <chr> <dbl>
#> 1 1 actor1 Actor 1 actor 0.5
#> 2 2 actor2 Actor 2 actor 1
#> 3 3 actor3 Actress 3 actor 0.5
#> 4 4 statement1 The fascists will lose statement NaN
#> 5 5 statement2 Owls are great statement NaN
#> #
#> # Edge Data: 7 × 4
#> from to stance timestamp
#> <int> <int> <chr> <date>
#> 1 1 4 support 2012-02-01
#> 2 1 4 support 2013-02-01
#> 3 1 5 irrelevant 2014-02-01
#> # ℹ 4 more rows
Or maybe even better, you can use tidygraph verbs directly on discourse_graph objects:
disc_g |>
tidygraph::activate(nodes) |>
dplyr::filter(mode == "actor") |>
tidygraph::activate(edges) |>
dplyr::filter(stance == "support")
#> ----------------------------------------
#> A discourse graph with 3 actors and 2
#> statements
#> ----------------------------------------
#> # A tbl_graph: 3 nodes and 0 edges
#> #
#> # A rooted forest with 3 trees
#> #
#> # Edge Data: 0 × 4 (active)
#> # ℹ 4 variables: from <int>, to <int>, stance <chr>, timestamp <date>
#> #
#> # Node Data: 3 × 4
#> nodeid name label mode
#> <int> <chr> <chr> <chr>
#> 1 1 actor1 Actor 1 actor
#> 2 2 actor2 Actor 2 actor
#> 3 3 actor3 Actress 3 actor
disc_g |>
tidygraph::activate(nodes) |>
dplyr::mutate(
degree = tidygraph::centrality_degree(),
name_upper = toupper(name)
)
#> ----------------------------------------
#> A discourse graph with 3 actors and 2
#> statements
#> ----------------------------------------
#> # A tbl_graph: 5 nodes and 7 edges
#> #
#> # A directed acyclic multigraph with 1 component
#> #
#> # Node Data: 5 × 6 (active)
#> nodeid name label mode degree name_upper
#> <int> <chr> <chr> <chr> <dbl> <chr>
#> 1 1 actor1 Actor 1 actor 4 ACTOR1
#> 2 2 actor2 Actor 2 actor 1 ACTOR2
#> 3 3 actor3 Actress 3 actor 2 ACTOR3
#> 4 4 statement1 The fascists will lose statement 0 STATEMENT1
#> 5 5 statement2 Owls are great statement 0 STATEMENT2
#> #
#> # Edge Data: 7 × 4
#> from to stance timestamp
#> <int> <int> <chr> <date>
#> 1 1 4 support 2012-02-01
#> 2 1 4 support 2013-02-01
#> 3 1 5 irrelevant 2014-02-01
#> # ℹ 4 more rows
Aggregating a discourse graph here means combining stance edges to weighted edges over time, possibly also keeping only the most prevalent category.
disc_g |> aggregate_discourse_graph()
#> ----------------------------------------
#> A aggregated discourse graph with 3 actors and 2
#> statements
#> ----------------------------------------
#> # A tbl_graph: 5 nodes and 5 edges
#> #
#> # A directed acyclic simple graph with 1 component
#> #
#> # Node Data: 5 × 4 (active)
#> nodeid name label mode
#> <int> <chr> <chr> <chr>
#> 1 1 actor1 Actor 1 actor
#> 2 2 actor2 Actor 2 actor
#> 3 3 actor3 Actress 3 actor
#> 4 4 statement1 The fascists will lose statement
#> 5 5 statement2 Owls are great statement
#> #
#> # Edge Data: 5 × 5
#> from to stance n_stances timestamp
#> <int> <int> <chr> <int> <date>
#> 1 1 4 support 3 2016-03-01
#> 2 1 5 irrelevant 1 2016-03-01
#> 3 2 5 opposition 1 2016-03-01
#> # ℹ 2 more rows
disc_g |> aggregate_discourse_graph(keep_only_highest = TRUE) |> plot()
start_date <- "2012-01-01"
end_date <- "2014-06-06"
disc_g |>
time_slice_graph(start_date = start_date, end_date = end_date)
#> ----------------------------------------
#> A discourse graph with 3 actors and 2
#> statements
#> ----------------------------------------
#> # A tbl_graph: 5 nodes and 4 edges
#> #
#> # A directed acyclic multigraph with 3 components
#> #
#> # Node Data: 5 × 4 (active)
#> nodeid name label mode
#> <int> <chr> <chr> <chr>
#> 1 1 actor1 Actor 1 actor
#> 2 2 actor2 Actor 2 actor
#> 3 3 actor3 Actress 3 actor
#> 4 4 statement1 The fascists will lose statement
#> 5 5 statement2 Owls are great statement
#> #
#> # Edge Data: 4 × 4
#> from to stance timestamp
#> <int> <int> <chr> <date>
#> 1 1 4 support 2012-02-01
#> 2 1 4 support 2013-02-01
#> 3 1 5 irrelevant 2014-02-01
#> # ℹ 1 more row
disc_g |>
time_slice_graph(start_date = start_date, end_date = end_date) |>
plot()
You can also create a list of time sliced graphs directly.
date_range <- c(start_date, end_date)
time_window <- months(48)
disc_g |>
time_sliced_graph_list(time_window = time_window,
date_range = date_range,
step_interval = "year")
#> $`2012-01-01`
#> ----------------------------------------
#> A discourse graph with 3 actors and 2
#> statements
#> ----------------------------------------
#> # A tbl_graph: 5 nodes and 2 edges
#> #
#> # A directed acyclic multigraph with 4 components
#> #
#> # Node Data: 5 × 4 (active)
#> nodeid name label mode
#> <int> <chr> <chr> <chr>
#> 1 1 actor1 Actor 1 actor
#> 2 2 actor2 Actor 2 actor
#> 3 3 actor3 Actress 3 actor
#> 4 4 statement1 The fascists will lose statement
#> 5 5 statement2 Owls are great statement
#> #
#> # Edge Data: 2 × 4
#> from to stance timestamp
#> <int> <int> <chr> <date>
#> 1 1 4 support 2012-02-01
#> 2 1 4 support 2013-02-01
#>
#> $`2013-01-01`
#> ----------------------------------------
#> A discourse graph with 3 actors and 2
#> statements
#> ----------------------------------------
#> # A tbl_graph: 5 nodes and 4 edges
#> #
#> # A directed acyclic multigraph with 3 components
#> #
#> # Node Data: 5 × 4 (active)
#> nodeid name label mode
#> <int> <chr> <chr> <chr>
#> 1 1 actor1 Actor 1 actor
#> 2 2 actor2 Actor 2 actor
#> 3 3 actor3 Actress 3 actor
#> 4 4 statement1 The fascists will lose statement
#> 5 5 statement2 Owls are great statement
#> #
#> # Edge Data: 4 × 4
#> from to stance timestamp
#> <int> <int> <chr> <date>
#> 1 1 4 support 2012-02-01
#> 2 1 4 support 2013-02-01
#> 3 1 5 irrelevant 2014-02-01
#> # ℹ 1 more row
#>
#> $`2014-01-01`
#> ----------------------------------------
#> A discourse graph with 3 actors and 2
#> statements
#> ----------------------------------------
#> # A tbl_graph: 5 nodes and 4 edges
#> #
#> # A directed acyclic multigraph with 3 components
#> #
#> # Node Data: 5 × 4 (active)
#> nodeid name label mode
#> <int> <chr> <chr> <chr>
#> 1 1 actor1 Actor 1 actor
#> 2 2 actor2 Actor 2 actor
#> 3 3 actor3 Actress 3 actor
#> 4 4 statement1 The fascists will lose statement
#> 5 5 statement2 Owls are great statement
#> #
#> # Edge Data: 4 × 4
#> from to stance timestamp
#> <int> <int> <chr> <date>
#> 1 1 4 support 2012-02-01
#> 2 1 4 support 2013-02-01
#> 3 1 5 irrelevant 2014-02-01
#> # ℹ 1 more row
You can explodes all statement nodes into all existing combinations of statements and stances. Easier to understand with an illustration:
disc_g |> explode_graph() |> plot()
