R/part_of_speech.R

In UBC-MDS/nlpsummarizer: nlpsummarizer - Finds Summary of Text data

# author : Karlos Muradyan
# date : 26 Feb 2020

#' This function generates statistics about the proportions of following
#' parts of speech in the given column:
#'     - verbs
#'     - prepositions
#'     - adjectives
#'     - nouns
#'     - articles
#'     - etc.
#'
#' @param df_col a list of values in data.frame column with textual data
#' @param model_path a string containing the path to the pretrained POS model.
#'             if provided, the function will look at it there, otherwise, the default place
#'	       is considered to be data/ folder. If the model doesn't exist, it will be downloaded.
#'
#' @return data.frame with one row containing columns verbs, prepositions,
#'	 adjectives, nouns and articles.
#'
#' @export
#'
#' @examples
#' ex <- data.frame(text_col =  c('Today is a beautiful Monday'))
#' get_part_of_speech(ex$text_col)


get_part_of_speech <- function(df_col, model_path = './data/english-ewt-ud-2.4-190531.udpipe'){
	if (!(class(df_col) == "character" || class(df_col) == "factor")){
		stop('The function should get input character or factor columns of dataframe')
	}

	print('If pre-trained model does not exist, it will be downloaded and put in ./data/ folder')
	if (!file.exists(model_path)){
		model_path <- udpipe::udpipe_download_model(language = "english", model_dir = './data')$file_model
	}
	udmodel <- udpipe::udpipe_load_model(file = model_path)
	tagged_text <- udpipe::udpipe_annotate(udmodel, df_col)
	resulting_df <- as.data.frame(tagged_text)

	proportions <- table(resulting_df$upos)/nrow(resulting_df)
	results <- as.data.frame(t(as.matrix(data.frame(proportions))))
	names(results) <- as.character(unlist(results[1,]))
	results <- results[-1,]
	return(results)
}