R/nodeindex.R

In csqsiew/langnetr: Creates a language network based on similarity

#' Replaces node IDs with their labels.
#'
#' @param langnet An igraph object created using the \code{tolangnet} function.
#' @param wordlist A list of words. Must be a character vector.
#' @return An igraph object of the language network with updated labels.
#' @examples
#' somewords <- c('cat', 'bat', 'cap', 'cape')
#' somewordsnet <- tolangnet(somewords)
#' plot(somewordsnet) #plots the graph
#'
#' somewordsnet.labelled <- nodeindex(somewordsnet, somewords)
#' plot(somewordsnet.labelled) #notice that this network has labels...
#' # note that the generated IDs are dependent on the ordering in the \code{wordlist}.

nodeindex <- function(langnet, wordlist) {                       # network created from the wordlist

  oldnames <- as.data.frame(as.numeric(igraph::V(langnet)$name)) # set up the node ids from the language network
  colnames(oldnames) <- 'nodeid'

  newnames <- as.data.frame(wordlist) # set up the wordlist for merging
  colnames(newnames) <- 'label'
  newnames$nodeid <- c(1:nrow(newnames))

  oldnames <- dplyr::left_join(oldnames, newnames, by = 'nodeid') # merge node ids and labels

  igraph::V(langnet)$name <- as.character(oldnames$label) # replace old names (ids) with new names (labels)

  return(langnet) #returns the igraph network with updated labels
}

# You can output to a .net file for further analysis, in gephi or pajek
# write.graph(test_graph, file='test_graph.net', format='pajek')
# BUT doesn't perserve labels - even node ids get completely shuffled
# Have to figure out a more robust way of export graphs.