word_associate: Find Associated Words.

In trinker/qdap2: Bridging the gap between qualitative data and quantitative analysis

Description

Find words associated with a given word(s) or a phrase(s). Results can be output as a network graph and/or wordcloud.

Usage

  word_associate(text.var, grouping.var = NULL,
    match.string, text.unit = "sentence",
    extra.terms = NULL, target.exclude = NULL,
    stopwords = NULL, network.plot = FALSE,
    wordcloud = FALSE, cloud.colors = c("black", "gray55"),
    title.color = "blue", nw.label.cex = 0.8,
    title.padj = -4.5, nw.label.colors = NULL,
    nw.layout = NULL, nw.edge.color = "gray90",
    nw.label.proportional = TRUE, nw.title.padj = NULL,
    nw.title.location = NULL, title.font = NULL,
    title.cex = NULL, nw.edge.curved = TRUE,
    cloud.legend = NULL, cloud.legend.cex = 0.8,
    cloud.legend.location = c(-0.03, 1.03),
    nw.legend = NULL, nw.legend.cex = 0.8,
    nw.legend.location = c(-1.54, 1.41),
    legend.override = FALSE, char2space = "~~", ...)

Arguments

text.var	The text variable.
grouping.var	The grouping variables. Default NULL generates one output for all text. Also takes a single grouping variable or a list of 1 or more grouping variables.
match.string	A list of vectors or vector of terms to associate in the text.
text.unit	The text unit (either "sentence" or "tot". This argument determines what unit to find the match string words within. For example if "sentence" is chosen the function pulls all text for sentences the match string terms are found in.
extra.terms	Other terms to color beyond the match string.
target.exclude	A vector of words to exclude from the match.string.
stopwords	Words to exclude from the analysis.
network.plot	logical. If TRUE plots a network plot of the words.
wordcloud	logical. If TRUE plots a wordcloud plot of the words.
cloud.colors	A vector of colors equal to the length of match.string +1.
title.color	A character vector of length one corresponding to the color of the title.
nw.label.cex	The magnification to be used for network plot labels relative to the current setting of cex. Default is .8.
title.padj	Adjustment for the title. For strings parallel to the axes, padj = 0 means right or top alignment, and padj = 1 means left or bottom alignment.
nw.label.colors	A vector of colors equal to the length of match.string +1.
nw.layout	layout types supported by igraph. See layout.
nw.edge.color	A character vector of length one corresponding to the color of the plot edges.
nw.label.proportional	logical. If TRUE scales the network plots across grouping.var to allow plot to plot comparisons.
nw.title.padj	Adjustment for the network plot title. For strings parallel to the axes, padj = 0 means right or top alignment, and padj = 1 means left or bottom alignment.
nw.title.location	On which side of the network plot (1=bottom, 2=left, 3=top, 4=right).
title.font	The font family of the cloud title.
title.cex	Character expansion factor for the title. NULL and NA are equivalent to 1.0.
nw.edge.curved	logical. If TRUE edges will be curved rather than straight paths.
cloud.legend	A character vector of names corresponding to the number of vectors in match.string. Both nw.legend and cloud.legend can be set separately; or one may be set and by default the other will assume those legend labels. If the user does not desire this behavior use the legend.override argument.
cloud.legend.cex	Character expansion factor for the wordcloud legend. NULL and NA are equivalent to 1.0.
cloud.legend.location	The x and y co-ordinates to be used to position the wordcloud legend. The location may also be specified by setting x to a single keyword from the list "bottomright", "bottom", "bottomleft", "left", "topleft", "top", "topright", "right" and "center". This places the legend on the inside of the plot frame at the given location.
nw.legend	A character vector of names corresponding to the number of vectors in match.string. Both nw.legend and cloud.legend can be set separately; or one may be set and by default the other will assume those legend labels. If the user does not desire this behavior use the legend.override argument.
nw.legend.cex	Character expansion factor for the network plot legend. NULL and NA are equivalent to 1.0.
nw.legend.location	The x and y co-ordinates to be used to position the network plot legend. The location may also be specified by setting x to a single keyword from the list "bottomright", "bottom", "bottomleft", "left", "topleft", "top", "topright", "right" and "center". This places the legend on the inside of the plot frame at the given location.
legend.override	By default if legend labels are supplied to either cloud.legend or nw.legend may be set and if the other remains NULL it will assume the supplied vector to the previous legend argument. If this behavior is not desired legend.override should be set to TRUE.
char2space	Currently a road to nowhere. Eventually this will allow the retention of characters as is allowed in trans.cloud already.
...	Other arguments supplied to trans.cloud.

Value

Returns a list:

word frequency matrices	Word frequency matrices for each grouping variable.
dialogue	A list of dataframes for each word list (each vector supplied to match.string) and a final dataframe of all combined text units that contain any match string.
match.terms	A list of vectors of word lists (each vector supplied to match.string).

Optionally, returns a word cloud and/or a network plot of the text unit containing the match.string terms.

See Also

trans.cloud, word.network.plot, wordcloud, graph.adjacency

Examples

ms <- c(" I", "you")
et <- c(" it", " no")
word_associate(DATA2$state, DATA2$person, match.string = ms,
    wordcloud = TRUE,  proportional = TRUE,
    network.plot = TRUE,  nw.label.proportional = TRUE, extra.terms = et,
    cloud.legend =c("A", "B", "C", "D"),
    title.color = "blue", cloud.colors = c("red", "blue", "purple", "gray70"))

#======================================
#Note: You don't have to name the vectors in the lists but I do for clarity
ms <- list(
    list1 = c(" I ", " you"),
    list2 = c(" wh")
)

et <- list(
    B = c(" the", " on"),
    C = c(" it", " no")
)

word_associate(DATA2$state, DATA2$person, match.string = ms,
    wordcloud = TRUE,  proportional = TRUE,
    network.plot = TRUE,  nw.label.proportional = TRUE, extra.terms = et,
    cloud.legend =c("A", "B", "C", "D"),
    title.color = "blue", cloud.colors = c("red", "blue", "purple", "gray70"))

word_associate(DATA2$state, list(DATA2$day, DATA2$person), match.string = ms)

#======================================
m <- list(
    A1 = c("you", "in"), #list 1
    A2 = c(" wh")        #list 2
)

n <- list(
    B = c(" the", " on"),
    C = c(" it", " no")
)

word_associate(DATA2$state, list(DATA2$day, DATA2$person), match.string = m)
word_associate(raj.act.1$dialogue, list(raj.act.1$person), match.string = m)
(out <- with(mraja1spl, word_associate(dialogue, list(fam.aff, sex), match.string = m)))
names(out)
lapply(out$dialogue, htruncdf, n = 20, w = 20)
out$cap.f

#======================================
DATA2$state2 <- space_fill(DATA2$state, c("is fun", "too fun"))

ms <- list(
    list1 = c(" I ", " you", "is fun", "too fun"),
    list2 = c(" wh")
)

et <- list(
    B = c(" the", " on"),
    C = c(" it", " no")
)

word_associate(DATA2$state2, DATA2$person, match.string = ms,
    wordcloud = TRUE,  proportional = TRUE,
    network.plot = TRUE,  nw.label.proportional = TRUE, extra.terms = et,
    cloud.legend =c("A", "B", "C", "D"),
    title.color = "blue", cloud.colors = c("red", "blue", "purple", "gray70"))

DATA2 <- qdap::DATA2

trinker/qdap2 documentation built on May 31, 2019, 9:47 p.m.