README.md

In crcox/netbuildr: Build adjacency matrices

netbuildr

Build associative networks from multiple data modalities, including: * word association norms * natural language corpora

The networks produced are represented as square adjacency matrices.

How to install

netbuildr is not on CRAN, but can be installed directly from github using the devtools package:

devtools::install_github("crcox/netbuildr")

How to use

Build network from text split across multiple documents

In the following, single letters are substituted for words. Each vector contained in the list tokens represents a document.

The window is forward looking, and directed connections are drawn from the first word in a window to the other words in the window. If a list of types are provided, only cooccurrences among the provided types are counted. The resulting matrix can be interpretted as a directed weighted graph, and can be made unweighted by applying a threshold.

tokens <- list(
    c("a", "g", "f", "b", "a", "e", "c", "d", "b", "a"),
    c("a", "h", "i", "b", "a", "j", "c", "k", "b", "a")
)
types <- c("a", "b", "c")
window_size <- 3
create_cooccurrence_matrix(tokens, window_size)
#   a b c d e f g h i j k
# a 0 0 2 0 1 1 1 1 1 1 0
# b 4 0 0 0 1 0 0 0 0 1 0
# c 0 2 0 1 0 0 0 0 0 0 1
# d 1 1 0 0 0 0 0 0 0 0 0
# e 0 0 1 1 0 0 0 0 0 0 0
# f 1 1 0 0 0 0 0 0 0 0 0
# g 0 1 0 0 0 1 0 0 0 0 0
# h 0 1 0 0 0 0 0 0 1 0 0
# i 1 1 0 0 0 0 0 0 0 0 0
# j 0 0 1 0 0 0 0 0 0 0 1
# k 1 1 0 0 0 0 0 0 0 0 0


create_cooccurrence_matrix(tokens, window_size, types)
#   a b c
# a 0 0 2
# b 4 0 0
# c 0 2 0

Build network from association data

In an association task, participants are asked to respond with the first words that come to mind when presented with a cue. When building a network, the cue words represent the nodes. A connection is drawn between two nodes when one is produced as a response when cued by the other.

In the example, there are three unique cues, so the network will have three nodes. The function returns a logical matrix.

cues    <- c(1,1,1,1,2,2,2,2,3,3,3,3)
responses <- c(2,4,5,6,3,4,5,6,1,4,5,6)
create_unweighted_network(cues, responses)
#       1     2     3
# 1 FALSE  TRUE FALSE
# 2 FALSE FALSE  TRUE
# 3  TRUE FALSE FALSE

crcox/netbuildr documentation built on Dec. 19, 2021, 6:19 p.m.