media_network_german_bundestag

library(rtweet)
library(tidyverse)
library(ggraph)
library(tidygraph)
library(ggrepel)

# get all the members from the bundestag from each faction, use lists provided by @pollytix_gmbh
mdb_list <- lists_members("973527951173193728") %>% mutate(party = "AfD") %>%
  bind_rows(lists_members("973614171886686209") %>% mutate(party = "CDU")) %>%
  bind_rows(lists_members("973520347634028545") %>% mutate(party = "CSU")) %>%
  bind_rows(lists_members("973546510456705024") %>% mutate(party = "B90/Die Grünen")) %>%
  bind_rows(lists_members("973561044709838849") %>% mutate(party = "Die Linke")) %>%
  bind_rows(lists_members("973485249001934848") %>% mutate(party = "FDP")) %>%
  bind_rows(lists_members("973582861361143809") %>% mutate(party = "SPD"))

party_color <- c(
  "AfD" = "#1a9fde",
  "SPD" = "#e10b1f", 
  "CDU" = "#565656", 
  "CSU" = "#727272", 
  "B90/Die Grünen" = "#499533", 
  "Die Linke" = "#bc3475", 
  "FDP" = "#e5d82d"
)

# get the last 3200 tweets of every user in that list
mdb_tweets <- get_timelines(mdb_list$user_id, n = 3200)

# not all profiles are public, lets check how many we got
mdb_tweets %>% distinct(user_id) %>% nrow()
# 538 from 547, that's okay (some may didn't tweet)

mdb_tweets %>% nrow()
# 976 508 tweets, keep in mind: max 3200 tweets per account

# extract certain german media
media_data <- tibble::tribble(
           ~media_regex,                     ~label_media,            ~media_short,
              "sz\\.de",            "Süddeutsche Zeitung",            "sz.de",
    "sueddeutsche\\.de",            "Süddeutsche Zeitung",            "sueddeutsche.de",
      "tagesschau\\.de",                     "Tagesschau",            "tagesschau.de",
            "bild\\.de",                           "Bild",            "bild.de",
            "welt\\.de",                           "Welt",            "welt.de",
           "focus\\.de",                          "Focus",            "focus.de",
         "spiegel\\.de",                        "Spiegel",            "spiegel.de",
#          "stern\\.de",                          "Stern",            "stern.de",
#      "jungewelt\\.de",                     "Junge Welt",            "jungewelt.de",
            "zeit\\.de",                    "Zeit Online",            "zeit.de",
#  "jungefreiheit\\.de",                 "Junge Freiheit",            "jungefreiheit.de",
#      "jungle\\.world",                   "Jungle World",            "jungle.world",
             "taz\\.de",                            "taz",            "taz.de",
#        "freitag\\.de",                    "Der Freitag",            "freitag.de",
  "compact-online\\.de",                "Compact Magazin",            "compact-online.de",
# "tichyseinblick\\.de",                "Tichys Einblick",            "tichyseinblick.de",
    "reitschuster\\.de",                   "Reitschuster",            "reitschuster.de",
            "faz\\.net", "Frankfurter Allgemeine Zeitung",            "faz.net",
    "handelsblatt\\.de",                   "Handelsblatt",            "handelsblatt.de",
      "nd-aktuell\\.de",              "neues deutschland",            "nd-aktuell.de",
  ) 


# count the media mentions per user_id
mdb_media_mention <- mdb_tweets %>%
  select(user_id, screen_name, name, urls_expanded_url) %>%
  unnest(urls_expanded_url) %>% 
  mutate(media = str_extract(urls_expanded_url, paste(media_data$media_regex, collapse = "|"))) %>%
  filter(!is.na(media)) %>%
  left_join(media_data, by = c("media" = "media_short")) %>%
  count(user_id, label_media)

# set random labels for mdbs (used for later)
set.seed(4465) # random seed, just for reproduction
mdb_labels <- mdb_list %>%
  group_by(party) %>%
  sample_n(2) %>%
  mutate(label_mdb = paste0(str_remove(name, ",.*$|MdB.*$") %>% str_trim(), "\n", party)) %>%
  ungroup() %>%
  select(user_id, label_mdb)

media_mention_total <- mdb_media_mention %>% 
  group_by(label_media) %>%
  summarise(n = sum(n)) %>%
  arrange(desc(n)) %>%
  mutate(user_id = label_media) # needed for joining later


graph_dat <- mdb_media_mention %>%
  as_tbl_graph() %>%
  activate(nodes) %>%
  rename(user_id = name) %>% # as_tbl_graph automatic renaming to name
  left_join(mdb_list %>% 
              left_join(mdb_labels, by = "user_id"),
            by = "user_id") %>%
  left_join(media_mention_total %>% mutate(type = "media"), by = "user_id") %>%
  mutate(label = if_else(is.na(label_media), label_mdb, label_media)) %>%
  mutate(party = if_else(is.na(label_media), party, "media")) %>%
  mutate(n = if_else(is.na(n), 10L, n)) # used as minimal size for later

ggraph(graph_dat, layout = "sugiyama") +
  geom_edge_link(alpha = 0.01) +
  geom_node_point(aes(size = n, color = party), alpha = 1) +
  geom_node_label(aes(label = label),
                  repel = TRUE, size = 3, segment.color = "#474747",
                  alpha = 0.8,
                  direction = "y", 
                  min.segment.length = 0,
                  position = position_nudge_repel(y = 0.1),
                  max.overlaps = 20) + 
  scale_color_manual(values = party_color) +
  theme_void() + theme(legend.position="none")