Text Filters

Get or specify the process by which text gets transformed into a sequence of tokens or sentences.

text_filter(x = NULL, ...)
text_filter(x) <- value

# S3 method for corpus_text
text_filter(x = NULL, ...)

# S3 method for data.frame
text_filter(x = NULL, ...)

# S3 method for default
text_filter(x = NULL, ...,
            map_case = TRUE, map_quote = TRUE,
            remove_ignorable = TRUE,
            combine = NULL,
            stemmer = NULL, stem_dropped = FALSE,
            stem_except = NULL,
            drop_letter = FALSE, drop_number = FALSE,
            drop_punct = FALSE, drop_symbol = FALSE,
            drop = NULL, drop_except = NULL,
            connector = "_",
            sent_crlf = FALSE,
            sent_suppress = corpus::abbreviations_en)

Arguments

x	text or corpus object.
value	text filter object, or NULL for the default.
...	further arguments passed to or from other methods.
map_case	a logical value indicating whether to apply Unicode case mapping to the text. For most languages, this transformation changes uppercase characters to their lowercase equivalents.
map_quote	a logical value indicating whether to replace curly single quotes and other Unicode apostrophe characters with ASCII apostrophe (U+0027).
remove_ignorable	a logical value indicating whether to remove Unicode "default ignorable" characters like zero-width spaces and soft hyphens.
combine	a character vector of multi-word phrases to combine, or NULL; see ‘Combining words’.
stemmer	a character value giving the name of a Snowball stemming algorithm (see stem_snowball for choices), a custom stemming function, or NULL to leave words unchanged.
stem_dropped	a logical value indicating whether to stem words in the "drop" list.
stem_except	a character vector of exception words to exempt from stemming, or NULL. If left unspecified, stem_except is set equal to the drop argument.
drop_letter	a logical value indicating whether to replace "letter" tokens (cased letters, kana, ideographic, letter-like numeric characters and other letters) with NA.
drop_number	a logical value indicating whether to replace "number" tokens (decimal digits, words appearing to be numbers, and other numeric characters) with NA.
drop_punct	a logical value indicating whether to replace "punct" tokens (punctuation) with NA.
drop_symbol	a logical value indicating whether to replace "symbol" tokens (emoji, math, currency, URLs, and other symbols) with NA.
drop	a character vector of types to replace with NA, or NULL.
drop_except	a character of types to exempt from the drop rules specified by the drop_letter, drop_number, drop_punct, drop_symbol, and drop arguments, or NULL.
connector	a character to use as a connector in lieu of white space for types that stem to multi-word phrases.
sent_crlf	a logical value indicating whether to break sentences on carriage returns or line feeds.
sent_suppress	a character vector of sentence break suppressions.

Details

The set of properties in a text filter determine the tokenization and sentence breaking rules. See the documentation for text_tokens and text_split for details on the tokenization process.

Value

text_filter retrieves an objects text filter, optionally with modifications to some of its properties.

text_filter<- sets an object's text filter. Setting the text filter on a character object is not allowed; the object must have type "corpus_text" or be a data frame with a "text" column of type "corpus_text".

See also

as_corpus_text, text_tokens, text_split, abbreviations, stopwords.

Examples

# text filter with default options set
text_filter()
#> Text filter with the following options:
#> 
#>     map_case: TRUE
#>     map_quote: TRUE
#>     remove_ignorable: TRUE
#>     combine: NULL
#>     stemmer: NULL
#>     stem_dropped: FALSE
#>     stem_except: NULL
#>     drop_letter: FALSE
#>     drop_number: FALSE
#>     drop_punct: FALSE
#>     drop_symbol: FALSE
#>     drop: NULL
#>     drop_except: NULL
#>     connector: _
#>     sent_crlf: FALSE
#>     sent_suppress:  chr [1:155] "A." "A.D." "a.m." "A.M." "A.S." "AA." ...

# specify some options but leave others unchanged
f <- text_filter(map_case = FALSE, drop = stopwords_en)

# set the text filter property
x <- as_corpus_text(c("Marnie the Dog is #1 on the internet."))
text_filter(x) <- f
text_tokens(x) # by default, uses x's text_filter to tokenize
#> [[1]]
#> [1] "Marnie"   "Dog"      "#"        "1"        "internet" "."       
#> 

# change a filter property
f2 <- text_filter(x, map_case = TRUE)
# equivalent to:
# f2 <- text_filter(x)
# f2$map_case <- TRUE

text_tokens(x, f2) # override text_filter(x)
#> [[1]]
#> [1] "marnie"   "dog"      "#"        "1"        "internet" "."       
#> 

# setting text_filter on a data frame is allowed if it has a
# column names "text" of type "corpus_text"
d <- data.frame(text = x)
text_filter(d) <- f2
text_tokens(d)
#> [[1]]
#> [1] "marnie"   "dog"      "#"        "1"        "internet" "."       
#> 

# but you can't set text filters on character objects
y <- "hello world"
if (FALSE) text_filter(y) <- f2 # gives an error

d2 <- data.frame(text = "hello world", stringsAsFactors = FALSE)
if (FALSE) text_filter(d2) <- f2 # gives an error