backbone_filter: Extract the backbone of a network.

In corpustools: Managing, Querying and Analyzing Tokenized Text

Extract the backbone of a network.

Description

Based on the following paper: Serrano, M. A., Boguna, M., & Vespignani, A. (2009). Extracting the multiscale backbone of complex weighted networks. Proceedings of the National Academy of Sciences, 106(16), 6483-6488.

Usage

backbone_filter(
  g,
  alpha = 0.05,
  direction = "none",
  delete_isolates = T,
  max_vertices = NULL,
  use_original_alpha = T,
  k_is_n = F
)

Arguments

g	A graph in the 'Igraph' format.
alpha	The threshold for the alpha. Can be interpreted similar to a p value (see paper for clarrification).
direction	direction = 'none' can be used for both directed and undirected networks, and is (supposed to be) the disparity filter proposed in Serrano et al. (2009) is used. By setting to 'in' or 'out', the alpha is only calculated for out or in edges. This is an experimental use of the backbone extraction (so beware!) but it seems a logical application.
delete_isolates	If TRUE, vertices with degree 0 (i.e. no edges) are deleted.
max_vertices	Optional. Set a maximum number of vertices for the network to be produced. The alpha is then automatically lowered to the point that only the given number of vertices remains connected (degree > 0). This can be usefull if the purpose is to make an interpretation friendly network. See e.g., http://jcom.sissa.it/archive/14/01/JCOM_1401_2015_A01
use_original_alpha	if max_vertices is not NULL, this determines whether the lower alpha for selecting the top vertices is also used as a threshold for the edges, or whether the original value given in the alpha parameter is used.
k_is_n	the disparity filter method for backbone extraction uses the number of existing edges (k) for each node, which can be arbitraty if there are many very weak ties, which is often the case in a co-occurence network. By setting k_is_n to TRUE, it is 'assumed' that all nodes are connected, which makes sense from a language model perspective (i.e. probability for co-occurence is never zero)

Value

A graph in the Igraph format

Examples

tc = create_tcorpus(sotu_texts, doc_column = 'id')
tc$preprocess('token','feature', remove_stopwords = TRUE, use_stemming = TRUE, min_docfreq = 10)

g = semnet_window(tc, 'feature', window.size = 10)
igraph::vcount(g)
igraph::ecount(g)
gb = backbone_filter(g, max_vertices = 100)
igraph::vcount(gb)
igraph::ecount(gb)
plot_semnet(gb)

corpustools documentation built on May 31, 2023, 8:45 p.m.