diskurs

⚠️ 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.

Installation

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'))

Basic functionality

library(diskurs)

Required data format

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

Creating a discourse graph

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
#> #
#> # A tibble: 5 × 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    
#> 4      4 statement1 The fascists will lose statement
#> 5      5 statement2 Owls are great         statement
#> #
#> # A tibble: 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 503552e DN-- 5 7 -- 
#> + attr: nodeid (v/n), name (v/c), label (v/c), mode (v/c), stance
#> | (e/c), timestamp (e/n)
#> + edges from 503552e (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.

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
#> #
#> # A tibble: 5 × 5
#>   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  
#> #
#> # A tibble: 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 graph

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
#> #
#> # A tibble: 5 × 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    
#> 4      4 statement1 The fascists will lose statement
#> 5      5 statement2 Owls are great         statement
#> #
#> # A tibble: 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()

Time slices of discourse graphs

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
#> #
#> # A tibble: 5 × 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    
#> 4      4 statement1 The fascists will lose statement
#> 5      5 statement2 Owls are great         statement
#> #
#> # A tibble: 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
#> #
#> # A tibble: 5 × 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    
#> 4      4 statement1 The fascists will lose statement
#> 5      5 statement2 Owls are great         statement
#> #
#> # A tibble: 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
#> #
#> # A tibble: 5 × 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    
#> 4      4 statement1 The fascists will lose statement
#> 5      5 statement2 Owls are great         statement
#> #
#> # A tibble: 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
#> #
#> # A tibble: 5 × 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    
#> 4      4 statement1 The fascists will lose statement
#> 5      5 statement2 Owls are great         statement
#> #
#> # A tibble: 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

Explode the graph

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()