A slopegraph in R for use with with the wordVectors package

slopegraph.R

library(ggplot2)
library(wordVectors)

slopegraph = function(
set1 = "RMP"
,
set2 = "genderless_RMP"
,
word = "she"
,
length=10
) {
  s1 = get(set1)
  s2 = get(set2)
  all_words = c(names(nearest_to(s1,s1[[word]],length)),names(nearest_to(s2,s2[[word]],length)))
  all_words = all_words[all_words!=word] # Who cares, this will always be 1.
  p01 = as.vector(cosineSimilarity(s1[[word]],s1))
  p02 = as.vector(cosineSimilarity(s2[[word]],s2))

  p1 = p01[match(all_words,rownames(s1))]
  p1Ranks = rank(-p01)[match(all_words,rownames(s1))]
  p2 = p02[match(all_words,rownames(s2))]
  p2Ranks = rank(-p02)[match(all_words,rownames(s2))]
  
  plottable = data.frame(word = as.vector(all_words),set1 = as.vector(p1),set2 = as.vector(p2),rank1=p1Ranks,rank2=p2Ranks)
  
  library(ggplot2)  
  ggplot(plottable) + geom_text(aes(y=set1,label=paste0(word," (",rank1,")")),hjust=1,x=1,alpha=.5) + 
    geom_text(aes(y=set2,label=paste0(word," (",rank2,")")),hjust=0,x=2,alpha=.5) + 
    geom_segment(aes(x=1,xend=2,y=set1,yend=set2)) + scale_x_continuous(limits=c(0,3)) +
    scale_y_continuous(paste0("Cosine similarity to ",word))+ 
    theme_bw() + 
    labs (title=paste0("Similarity to '", word,"' in the\nvectorspaces ", set1, " (left) and\n", set2, " (right)"))
}