term_stats: Term Statistics

Description

Tokenize a set of texts and tabulate the term occurrence statistics.

Usage

term_stats(x, filter = NULL, weights = NULL,
           ngrams = NULL, min_count = NULL, max_count = NULL,
           min_support = NULL, max_support = NULL, types = FALSE,
           subset, ...)

Arguments

a text vector to tokenize.

filter

if non-NULL, a text filter to to use instead of the default text filter for x.

weights

a numeric vector the same length of x assigning weights to each text, or NULL for unit weights.

ngrams

an integer vector of n-gram lengths to include, or NULL for length-1 n-grams only.

min_count

a numeric scalar giving the minimum term count to include in the output, or NULL for no minimum count.

max_count

a numeric scalar giving the maximum term count to include in the output, or NULL for no maximum count.

min_support

a numeric scalar giving the minimum term support to include in the output, or NULL for no minimum support.

max_support

a numeric scalar giving the maximum term support to include in the output, or NULL for no maximum support.

types

a logical value indicating whether to include columns for the types that make up the terms.

subset

logical expression indicating elements or rows to keep: missing values are taken as false.

…

additional properties to set on the text filter.

Value

A data frame with columns named term, count, and support, with one row for each appearing term. Rows are sorted in descending order according to support and then count, with ties broken lexicographically by term, using the character ordering determined by the current locale (see Comparison for details).

If types = TRUE, then the result also includes columns named type1, type2, etc. for the types that make up the term.

Details

term_stats tokenizes a set of texts and computes the occurrence counts and supports for each term. The ‘count’ is the number of occurrences of the term across all texts; the ‘support’ is the number of texts containing the term. If weights is non-NULL, then each token in text i increments the count for the corresponding term by weights[i]; otherwise, each appearance increments the count by one. Likewise, for non-NULL weights, an appearance of a term in text i increments its support by weight[i] (once, not for each occurrence in the text).

To include multi-type terms, specify the designed term lengths using the ngrams argument.

Examples

Run this code

# NOT RUN {
term_stats("A rose is a rose is a rose.")

# remove punctuation and English stop words
term_stats("A rose is a rose is a rose.",
           text_filter(drop_symbol = TRUE, drop = stopwords_en))

# weight the texts
term_stats(c("A rose is a rose is a rose.",
             "A Rose is red, a violet is blue!"),
           weights = c(100, 1))

# unigrams, bigrams, and trigrams
term_stats("A rose is a rose is a rose.", ngrams = 1:3)

# also include the type information
term_stats("A rose is a rose is a rose.", ngrams = 1:3, types = TRUE)
# }