Struct linfa_preprocessing::count_vectorization::CountVectorizer

pub struct CountVectorizer { /* fields omitted */ }

Count vectorizer: learns a vocabulary from a sequence of documents (or file paths) and maps each vocabulary entry to an integer value, producing a FittedCountVectorizer that can be used to count the occurrences of each vocabulary entry in any sequence of documents. Alternatively a user-specified vocabulary can be used for fitting.

Attributes

If a user-defined vocabulary is used for fitting then the following attributes will not be considered during the fitting phase but they will still be used by the FittedCountVectorizer to transform any text to be examined.

• split_regex: the regex espression used to split decuments into tokens. Defaults to r“\b\w\w+\b“, which selects “words”, using whitespaces and punctuation symbols as separators.
• convert_to_lowercase: if true, all documents used for fitting will be converted to lowercase. Defaults to true.
• n_gram_range: if set to (1,1) single tokens will be candidate vocabulary entries, if (2,2) then adjacent token pairs will be considered, if (1,2) then both single tokens and adjacent token pairs will be considered, and so on. The definition of token depends on the regex used fpr splitting the documents. The default value is (1,1).
• normalize: if true, all charachters in the documents used for fitting will be normalized according to unicode’s NFKD normalization. Defaults to true.
• document_frequency: specifies the minimum and maximum (relative) document frequencies that each vocabulary entry must satisfy. Defaults to (0., 1.) (i.e. 0% minimum and 100% maximum)
• stopwords: optional list of entries to be excluded from the generated vocabulary. Defaults to None

Implementations

impl CountVectorizer

pub fn convert_to_lowercase(self, convert_to_lowercase: bool) -> Self

If true, all documents used for fitting will be converted to lowercase.

pub fn split_regex(self, regex_str: &str) -> Self

Sets the regex espression used to split decuments into tokens

pub fn n_gram_range(self, min_n: usize, max_n: usize) -> Self

If set to (1,1) single tokens will be candidate vocabulary entries, if (2,2) then adjacent token pairs will be considered, if (1,2) then both single tokens and adjacent token pairs will be considered, and so on. The definition of token depends on the regex used fpr splitting the documents.

min_n should not be greater than max_n

pub fn normalize(self, normalize: bool) -> Self

If true, all charachters in the documents used for fitting will be normalized according to unicode’s NFKD normalization.

pub fn document_frequency(self, min_freq: f32, max_freq: f32) -> Self

Specifies the minimum and maximum (relative) document frequencies that each vocabulary entry must satisfy. min_freq and max_freq must lie in [0;1] and min_freq should not be greater than max_freq

pub fn stopwords<T: ToString>(self, stopwords: &[T]) -> Self

List of entries to be excluded from the generated vocabulary.

pub fn fit<T: ToString + Clone, D: Data<Elem = T>>(
    &self,
    x: &ArrayBase<D, Ix1>
) -> Result<FittedCountVectorizer>

Learns a vocabulary from the documents in x, according to the specified attributes and maps each vocabulary entry to an integer value, producing a FittedCountVectorizer.

Returns an error if:

• one of the n_gram boundaries is set to zero or the minimum value is greater than the maximum value
• if the minimum document frequency is greater than one or than the maximum frequency, or if the maximum frequency is
  smaller than zero
• if the regex expression for the split is invalid

pub fn fit_files<P: AsRef<Path>>(
    &self,
    input: &[P],
    encoding: EncodingRef,
    trap: DecoderTrap
) -> Result<FittedCountVectorizer>

Learns a vocabulary from the documents contained in the files in input, according to the specified attributes and maps each vocabulary entry to an integer value, producing a FittedCountVectorizer.

The files will be read using the specified encoding, and any sequence unrecognized by the encoding will be handled according to trap.

Returns an error if:

• one of the n_gram boundaries is set to zero or the minimum value is greater than the maximum value
• if the minimum document frequency is greater than one or than the maximum frequency, or if the maximum frequency is
  smaller than zero
• if the regex expression for the split is invalid
• if one of the files couldn’t be opened
• if the trap is strict and an unrecognized sequence is encountered in one of the files

pub fn fit_vocabulary<T: ToString>(
    &self,
    words: &[T]
) -> Result<FittedCountVectorizer>

Produces a FittedCountVectorizer with the input vocabulary. All struct attributes are ignored in the fitting but will be used by the FittedCountVectorizer to transform any text to be examined. As such this will return an error in the same cases as the fit method.

Trait Implementations

impl Default for CountVectorizer

fn default() -> Self

Returns the “default value” for a type.