stylo (version 0.7.5)

make.table.of.frequencies: Prepare a table of (relative) word frequencies

Description

Function that collects several frequency lists and combines them into a single frequency table. To this end a number of rearrangements inside particular lists are carried out. The table is produced using a reference list of words/features (passed as an argument).

Usage

make.table.of.frequencies(corpus, features, absent.sensitive = TRUE, 
                          relative = TRUE)

Arguments

corpus

textual data: either a corpus (represented as a list), or a single text (represented as a vector); the data have to be split into words (or other features, such as character n-grams or word pairs).

features

a vector containing a reference feature list that will be used to build the table of frequencies (it is assumed that the reference list contains the same type of features as the corpus list, e.g. words, character n-grams, word pairs, etc.; otherwise, an empty table will be build).

absent.sensitive

this optional argument is used to prevent building tables of words/features that never occur in the corpus. When switched on (default), variables containing 0 values across all samples, will be excluded. However, in some cases this is important to keep all the variables regardless of their values. This is e.g. the case when comparing two corpora: even if a given word did not occur in corpus A, it might be present in corpus B. In short: whenever you perform any analysis involving two or multiple sets of texts, switch this option to FALSE.

relative

when this argument is switched to TRUE (default), relative frequencies are computed instead of raw frequencies.

Author

Maciej Eder

See Also

load.corpus, load.corpus.and.parse

Examples

# to get frequencies of the words "a", "the" and "of" from a text:

sample.txt = txt.to.words("My father had a small estate 
             in Nottinghamshire: I was the third of five sons.")
make.table.of.frequencies(sample.txt, c("a", "the", "of"))



# to get a table of frequencies across several texts:

txt.1 = "Gallia est omnis divisa in partes tres, quarum unam incolunt 
    Belgae, aliam Aquitani, tertiam qui ipsorum lingua Celtae, nostra 
    Galli appellantur."
txt.2 = "Si quis antea, iudices, mirabatur quid esset quod, pro tantis 
    opibus rei publicae tantaque dignitate imperi, nequaquam satis multi 
    cives forti et magno animo invenirentur qui auderent se et salutem 
    suam in discrimen offerre pro statu civitatis et pro communi 
    libertate, ex hoc tempore miretur potius si quem bonum et fortem 
    civem viderit, quam si quem aut timidum aut sibi potius quam rei 
    publicae consulentem."
txt.3 = "Nam mores et instituta vitae resque domesticas ac familiaris 
    nos profecto et melius tuemur et lautius, rem vero publicam nostri 
    maiores certe melioribus temperaverunt et institutis et legibus."
my.corpus.raw = list(txt.1, txt.2, txt.3)
my.corpus.clean = lapply(my.corpus.raw, txt.to.words)
my.favorite.words = c("et", "in", "se", "rara", "avis")
make.table.of.frequencies(my.corpus.clean, my.favorite.words)


# to include all words in the reference list, no matter if they 
# occurred in the corpus:
make.table.of.frequencies(my.corpus.clean, my.favorite.words,
    absent.sensitive=FALSE)


# to prepare a table of frequencies of all the words represented in 
# a corpus, in descendent occurence order, one needs to make the frequency
# list first, via the function 'make.frequency.list'
complete.word.list = make.frequency.list(my.corpus.clean)
make.table.of.frequencies(my.corpus.clean, complete.word.list)


# to create a table of frequencies of word pairs (word 2-grams):
my.word.pairs = lapply(my.corpus.clean, txt.to.features, ngram.size=2)
make.table.of.frequencies(my.word.pairs, c("et legibus", "hoc tempore"))