Skip to contents

Force Directed Hierarchical Cluster Tree Network

forceClusTree.Rd

Generates a Force Directed Hierarchical Cluster Tree Network Graph for corpus objects.

Usage

forceClusTree(corpus, 
              remove_punct=TRUE,
              lang = "es",
              palette = c("#DD8D29","#E2D200","#46ACC8","#E58601","#B40F20"),
              groupvar = NULL,
              width="100%",
              height=580,
              link.width=2,
              maxRadius=30,
              BodyStrength=-10,
              include.text=FALSE,
              img.docvar=NULL,
              clust.method="euclidean",
              weight.scheme="logcount",
              elementId="chartdivclus")

Arguments

corpus

A quanteda corpus object.

remove_punct

Remove punctuation. The default is TRUE.

lang

The language for removing stopwords. The default is Spanish: "es".

palette

The color palette to be employed.

groupvar

Metadata variable associated to the texts on a quanteda corpus object indicating groups or more aggregated categories.

width

The width of the html panel. The default is 100%.

height

The height of the html panel in pixels. The default is 580.

link.width

The width of the links connecting nodes in the network. The default is 2.

maxRadius

The maximum radius of the nodes in the network. The default is 100.

BodyStrength

The attraction force of nodes in the network. The default is -10.

include.text

Logical. Indicates if the function should include the text in the tooltip. It should be used only with small texts. Large texts would collapse the visualization. The default is FALSE.

img.docvar

Indicates the name of the variable in the corpus metadata containing the url links to the images to be displayed in the tooltip. The default is NULL.

clust.method

Indicates the method for calculating the text distance. The options are: "euclidean" (default), "manhattan", "maximum", "canberra", and "minkowski".

weight.scheme

Indicates the method for calculating the weights for pondering the significance of words in each text. The options are: "logcount" (default), "count", "prop", "propmax", "boolean", "augmented", "logave". See quateda's dfm_weight for more information on how each scheme is calculated.

elementId

Name of the div element employed to contain the graph. It is useful when you use the same type of graph multiple times in the same markdown page, for instance. The default is "chartdivclus".

Details

The function generates a Force Directed Hierarchical Cluster Tree Network Graph for quanteda corpuses. It calculates the similarity among texts and organizes them according to a tree structure. The interactive traits of the chart allow users to explore these relationships with more depth.

Value

A chart representing the similarity of texts or the html source code generated.

Examples

if (FALSE) {
# Retrieve a corpus of text 
tx <- quanteda::data_corpus_inaugural

# Create the graph for the inaugural discourses
forceClusTree(tx, lang="en", maxRadius = 50)
}