Analyze and Visualize Word Correlation Networks

This function creates a word correlation network based on a document-feature matrix (dfm).

Usage

word_correlation_network(
  dfm_object,
  doc_var = NULL,
  common_term_n = 130,
  corr_n = 0.4,
  top_node_n = 40,
  nrows = 1,
  height = 1000,
  width = 900,
  node_label_size = 22,
  community_method = "leiden",
  node_size_by = "degree",
  node_color_by = "community"
)

Arguments

dfm_object

A quanteda document-feature matrix (dfm).

doc_var

A document-level metadata variable (default: NULL).

common_term_n

Minimum number of common terms for filtering terms (default: 130).

corr_n

Minimum correlation value for filtering terms (default: 0.4).

top_node_n

Number of top nodes to display (default: 40).

nrows

Number of rows to display in the table (default: 1).

height

The height of the resulting Plotly plot, in pixels (default: 1000).

width

The width of the resulting Plotly plot, in pixels (default: 900).

node_label_size

Maximum font size for node labels in pixels (default: 22).

community_method

Community detection method: "leiden" (default) or "louvain".

node_size_by

Node sizing method: "degree", "betweenness", "closeness", "eigenvector", or "fixed" (default: "degree").

node_color_by

Node coloring method: "community" or "centrality" (default: "community").

Value

A list containing the Plotly plot, a table, and a summary.

See also

Other semantic: analyze_document_clustering(), analyze_similarity_gaps(), calculate_clustering_metrics(), calculate_cross_similarity(), calculate_document_similarity(), calculate_similarity_robust(), cluster_embeddings(), cross_analysis_validation(), export_document_clustering(), extract_cross_category_similarities(), fit_semantic_model(), generate_cluster_labels(), generate_cluster_labels_auto(), generate_embeddings(), reduce_dimensions(), semantic_document_clustering(), semantic_similarity_analysis(), temporal_semantic_analysis(), validate_cross_models(), word_co_occurrence_network()

Examples

if (interactive()) {
  df <- TextAnalysisR::SpecialEduTech

  united_tbl <- TextAnalysisR::unite_cols(df, listed_vars = c("title", "abstract"))

  tokens <- TextAnalysisR::prep_texts(united_tbl, text_field = "united_texts")

  dfm_object <- quanteda::dfm(tokens)

  word_correlation_network_results <- TextAnalysisR::word_correlation_network(
                                      dfm_object,
                                      doc_var = NULL,
                                      common_term_n = 30,
                                      corr_n = 0.4,
                                      top_node_n = 40,
                                      nrows = 1,
                                      height = 1000,
                                      width = 900,
                                      community_method = "leiden")
  print(word_correlation_network_results$plot)
  print(word_correlation_network_results$table)
  print(word_correlation_network_results$summary)
}