In PolMine/polmineR.graph: gradget

Loading required libraries

library(polmineR)
library(data.table)
library(polmineR.graph)
library(igraph)
library(RColorBrewer)
library(three)

Configuration

termOfInterest <- "Ungleichheit"
corpus <- "PLPRBT"
sAttr <- "speaker_year" # for PLPRBT, would be "text_year" for many other corpora
pAttr <- "lemma"
use.regex <- TRUE
years <- as.character(1998:2015)
window.size <- 10L

threshold <- 3.84
min.count <- 3
graph.order = 1L


svg.width <- 800
svg.height <- 650
svg.margin <- 100
svg.fontSize <- 13
svg.edgeColor <- "lightgrey"

setwd("~/Lab/tmp/ungleichheit") # output html files will be stored here

helper function

makeSVG <- function(sAttrValue, label){

  print(year)

  # create partition
  defList <- as.list(setNames(sAttrValue, sAttr))
  P <- partition(corpus, def = defList, pAttribute = pAttr, regex = use.regex)

  coocTest <- cooccurrences(P, query = termOfInterest, pAttribute = pAttr)
  if (length(which(coocTest[["count_window"]] >= 5)) <= 5){
    msg <- "less than 5 statistically significant cooccurrences"
    warning(msg, year)
    return(NULL)
  }

  # create Cooccurrences object
  termsToDrop <- c(polmineR::punctuation, unlist(noise(pAttributes(P, pAttribute = pAttr))))
  Cooc <- Cooccurrences$new(partition = P, pAttribute = pAttr, window = window.size, drop = termsToDrop)
  Cooc$count()
  Cooc$trim(action = "drop", by.id = TRUE)
  Cooc$maths()

  # minimize by applying statistical threshold, reduction to token neighborhood
  CoocMin <- copy(Cooc)
  CoocMin$dt <- copy(Cooc$dt)
  CoocMin$dt <- CoocMin$dt[ll > threshold]
  CoocMin$dt <- CoocMin$dt[count_ab >= min.count]
  G <- CoocMin$as.igraph(as.undirected = TRUE)
  if (! termOfInterest %in% V(G)$name){
    warning("no statistignificantly reliable result for ", year)
    return(NULL)
  }
  Gmin <- make_ego_graph(G, order = graph.order, nodes = termOfInterest)[[1]]
  Gmin <- addCommunities(Gmin)
  Gmin <- addCoordinates(Gmin, layout = "kamada.kawai", dim = 2)

  # turn into SVG
  S <- SVG$new(Gmin)
  S$width = svg.width
  S$height = svg.height
  S$margin = svg.margin
  S$fontSize = svg.fontSize
  S$edgeColor = svg.edgeColor
  S$make()
  outfile <- S$store(filename = file.path(getwd(), paste(year, ".html", sep = "")))
  print(outfile)

}

Generate output

for (year in years) makeSVG(sAttrValue = year, label = year)

makeSVG(sAttrValue = as.character(1998:2008), label = "pre2008")
makeSVG(sAttrValue = as.character(2008:2015), label = "2008ff")

PolMine/polmineR.graph documentation built on April 27, 2020, 3:24 a.m.