rsyntax (version 0.1.4)

plot_tree: Create an igraph tree from a sentence

Description

Create an igraph tree from a token_index (as_tokenindex) or a data.frame that can be coerced to a tokenindex.

By default, all columns in the data are included as labels. This can be changes by using the ... argument.

Usage

plot_tree(
  tokens,
  ...,
  sentence_i = 1,
  doc_id = NULL,
  sentence = NULL,
  annotation = NULL,
  only_annotation = FALSE,
  pdf_file = NULL,
  allign_text = TRUE,
  ignore_rel = NULL,
  all_lower = FALSE,
  all_abbrev = NULL,
  textsize = 1,
  spacing = 1,
  use_color = TRUE,
  max_curve = 0.3,
  palette = grDevices::terrain.colors,
  rel_on_edge = F,
  pdf_viewer = FALSE,
  viewer_mode = TRUE,
  viewer_size = c(100, 100)
)

Arguments

tokens

A tokenIndex data.table, or any data.frame coercible with as_tokenindex. Can also be a corpustools tCorpus.

...

Optionally, select which columns to include as labels and how to present them. Can be quoted or unquoted names and expressions, using columns in the tokenIndex. For example, plot_tree(tokens, token, pos) will use the $token and $pos columns in tokens. You can also use expressions for easy controll of visulizations. For example: plot_tree(tokens, tolower(token), abbreviate(pos,1)). (note that abbreviate() is really usefull here)

sentence_i

By default, plot_tree uses the first sentence (sentence_i = 1) in the data. sentence_i can be changed to select other sentences by position (the i-th unique sentence in the data). Note that sentence_i does not refer to the values in the sentence column (for this use the sentence argument together with doc_id)

doc_id

Optionally, the document id can be specified. If so, sentence_i refers to the i-th sentence within the given document.

sentence

Optionally, the sentence id can be specified (note that sentence_i refers to the position). If sentence is given, doc_id has to be given as well.

annotation

Optionally, a column with an rsyntax annotation, to add boxes around the annotated nodes.

only_annotation

If annotation is given, only_annotation = TRUE will print only the nodes with annotations.

pdf_file

Directly save the plot as a pdf file

allign_text

If TRUE (default) allign text (the columns specified in ...) in a single horizontal line at the bottom, instead of following the different levels in the tree

ignore_rel

Optionally, a character vector with relation names that will not be shown in the tree

all_lower

If TRUE, make all text lowercase

all_abbrev

If an integer, abbreviate all text, with the number being the target number of characters.

textsize

A number to manually change the textsize. The function tries to set a suitable textsize for the plotting device, but if this goes wrong and now everything is broken and sad, you can multiply the textsize with the given number.

spacing

A number for scaling the distance between words (between 0 and infinity)

use_color

If true, use colors

max_curve

A number for controlling the allowed amount of curve in the edges.

palette

A function for creating a vector of n contiguous colors. See ?terrain.colors for standard functions and documentation

rel_on_edge

If TRUE, print relation label on edge instead of above the node

pdf_viewer

If TRUE, view the plot as a pdf. If no pdf_file is specified, the pdf will be saved to the temp folder

viewer_mode

By default, the plot is saved as a PNG embedded in a HTML and opened in the viewer. This hack makes it independent of the size of the plotting device and enables scrolling. By setting viewer_mode to False, the current plotting device is used.

viewer_size

A vector of length 2, that multiplies the width (first value) and height (second value) of the viewer_mode PNG

Value

plots a dependency tree.

Examples

# NOT RUN {
tokens = tokens_spacy[tokens_spacy$doc_id == 'text3',]

# }
# NOT RUN {
if (interactive()) plot_tree(tokens, token, pos)

## plot with annotations
direct = tquery(label = 'verb', pos = 'VERB', fill=FALSE,
                children(label = 'subject', relation = 'nsubj'),
                children(label = 'object', relation = 'dobj'))
passive = tquery(label = 'verb', pos = 'VERB', fill=FALSE,
                 children(label = 'subject', relation = 'agent'),
                 children(label = 'object', relation = 'nsubjpass'))
 
if (interactive()) {                
tokens %>%
   annotate_tqueries('clause', pas=passive, dir=direct) %>%
   plot_tree(token, pos, annotation='clause')
}
# }