hackathon.r

library(quanteda)
library(dplyr)
library(tidyr)
library(networkD3)

load(url("http://www.kenbenoit.net/files/presDebateCorpus2016seg.RData"))
candidates <- subset(presDebateCorpus2016seg, speakertype == 'candidate')
dfm <- dfm(candidates, groups = c("tag"), ngrams = 1:3,
           ignoredFeatures = c('people','go','going','will','know','think',
                               'country','get','applause','want','need',
                               'can','say','president','one','well','make',
                               'us','said',stopwords("english")),
           stem = TRUE)
dist <- as.matrix(similarity(dfm))
cd <- row.names(dist)
candidate_list <- c("RUBIO","CLINTON","SANDERS","TRUMP","CRUZ","BUSH","O'MALLEY")
dist <- as.data.frame(dist) %>% 
  mutate(cd = cd) %>% 
  gather(key = to, value = dist, -cd) %>% 
  rename(from = cd) %>% 
  mutate(dist = 1-dist) %>% 
  filter(dist !=0) %>% 
  filter(from %in% candidate_list,
         to %in% candidate_list)

nodes <- unique(c(dist$from,dist$to))
nodes <- data.frame(NodeID = nodes, stringsAsFactors = F)
party <- data.frame(candidates$documents$party,candidates$documents$tag) %>% 
  unique() %>% 
  rename(group = candidates.documents.party,
         NodeID = candidates.documents.tag)
nodes <- inner_join(nodes,party)
from <- data.frame(from = nodes$NodeID,
                            from_id = seq(0,nrow(nodes)-1))
to <- data.frame(to = nodes$NodeID,
                           to_id = seq(0,nrow(nodes)-1))
dist <- dist %>% 
  inner_join(from) %>% 
  inner_join(to) %>% 
  select(-from,-to) %>% 
  rename(from=from_id,to=to_id)

nodes$group <- as.character(nodes$group)
nodes$group[nodes$group == 'republican'] <- 'Republican'
nodes$group[nodes$group == 'democratic'] <- 'Democrat'

forceNetwork(Links = dist, Nodes = nodes,
             Source = "from", Target = "to",
             Value = "dist", NodeID = "NodeID",
             Group = "group", opacity = 0.8, 
             colourScale = JS("d3.scale.category10()"),
             zoom = TRUE, legend = TRUE,
             linkDistance = JS("function(d){return d.value * 200}"),
             linkWidth = 0.1)