Type: Package
Title: An Interface to the 'fastText' Library
Version: 2.1.0
Description: An interface to the 'fastText' library https://github.com/facebookresearch/fastText. The package can be used for text classification and to learn word vectors. An example how to use 'fastTextR' can be found in the 'README' file.
License: BSD_3_clause + file LICENSE
Imports: stats, graphics, Rcpp (≥ 0.12.4), slam
Suggests: knitr, rmarkdown
VignetteBuilder: knitr
LinkingTo: Rcpp
Encoding: UTF-8
RoxygenNote: 7.2.3
URL: https://github.com/EmilHvitfeldt/fastTextR
BugReports: https://github.com/EmilHvitfeldt/fastTextR/issues
NeedsCompilation: yes
Packaged: 2023-12-08 23:17:48 UTC; emilhvitfeldt
Author: Florian Schwendinger [aut], Emil Hvitfeldt ORCID iD [aut, cre]
Maintainer: Emil Hvitfeldt <emilhhvitfeldt@gmail.com>
Repository: CRAN
Date/Publication: 2023-12-09 00:40:09 UTC

Create a New FastText Model

Description

Create a new FastText model. The available methods are the same as the package functions but with out the prefix "ft_" and without the need to provide the model.

Usage

fasttext()

Examples

ft <- fasttext()

Get Analogies

Description

TODO

Usage

ft_analogies(model, word_triplets, k = 10L)

Arguments

model

an object inheriting from "fasttext".

word_triplets

a character vector of length string giving the word.

k

an integer giving the number of nearest neighbors to be returned.

Value

.

Examples

## Not run: 
ft_analogies(model, c("berlin", "germany", "france"), k = 6L)

## End(Not run)

Default Control Settings

Description

A auxiliary function for defining the control variables.

Usage

ft_control(
  loss = c("softmax", "hs", "ns"),
  learning_rate = 0.05,
  learn_update = 100L,
  word_vec_size = 100L,
  window_size = 5L,
  epoch = 5L,
  min_count = 5L,
  min_count_label = 0L,
  neg = 5L,
  max_len_ngram = 1L,
  nbuckets = 2000000L,
  min_ngram = 3L,
  max_ngram = 6L,
  nthreads = 1L,
  threshold = 1e-04,
  label = "__label__",
  verbose = 0,
  pretrained_vectors = "",
  output = "",
  save_output = FALSE,
  seed = 0L,
  qnorm = FALSE,
  retrain = FALSE,
  qout = FALSE,
  cutoff = 0L,
  dsub = 2L,
  autotune_validation_file = "",
  autotune_metric = "f1",
  autotune_predictions = 1L,
  autotune_duration = 300L,
  autotune_model_size = ""
)

Arguments

loss

a character string giving the name of the loss function allowed values are 'softmax', 'hs' and 'ns'.

learning_rate

a numeric giving the learning rate, the default value is 0.05.

learn_update

an integer giving after how many tokens the learning rate should be updated. The default value is 100L, which means the learning rate is updated every 100 tokens.

word_vec_size

an integer giving the length (size) of the word vectors.

window_size

an integer giving the size of the context window.

epoch

an integer giving the number of epochs.

min_count

an integer giving the minimal number of word occurences.

min_count_label

and integer giving the minimal number of label occurences.

neg

an integer giving how many negatives are sampled (only used if loss is "ns").

max_len_ngram

an integer giving the maximum length of ngrams used.

nbuckets

an integer giving the number of buckets.

min_ngram

an integer giving the minimal ngram length.

max_ngram

an integer giving the maximal ngram length.

nthreads

an integer giving the number of threads.

threshold

a numeric giving the sampling threshold.

label

a character string specifying the label prefix (default is '__label__').

verbose

an integer giving the verbosity level, the default value is 0L and shouldn't be changed since Rcpp::Rcout cann't handle the traffic.

pretrained_vectors

a character string giving the file path to the pretrained word vectors which are used for the supervised learning.

output

a character string giving the output file path.

save_output

a logical (default is FALSE)

seed

an integer

qnorm

a logical (default is FALSE)

retrain

a logical (default is FALSE)

qout

a logical (default is FALSE)

cutoff

an integer (default is 0L)

dsub

an integer (default is 2L)

autotune_validation_file

a character string

autotune_metric

a character string (default is "f1")

autotune_predictions

an integer (default is 1L)

autotune_duration

an integer (default is 300L)

autotune_model_size

a character string

Value

a list with the control variables.

Examples

ft_control(learning_rate=0.1)

Load Model

Description

Load a previously saved model from file.

Usage

ft_load(file)

Arguments

file

a character string giving the name of the file to be read in.

Value

an object inheriting from "fasttext".

Examples

## Not run: 
model <- ft_load("dbpedia.bin")

## End(Not run)

Get Nearest Neighbors

Description

TODO

Usage

ft_nearest_neighbors(model, word, k = 10L)

Arguments

model

an object inheriting from "fasttext".

word

a character string giving the word.

k

an integer giving the number of nearest neighbors to be returned.

Value

.

Examples

## Not run: 
ft_nearest_neighbors(model, "enviroment", k = 6L)

## End(Not run)

Normalize

Description

Applies normalization to a given text.

Usage

ft_normalize(txt)

Arguments

txt

a character vector to be normalized.

Value

a character vector.

Examples

## Not run: 
ft_normalize(some_text)

## End(Not run)

Write Model

Description

Write a previously saved model from file.

Usage

ft_save(model, file, what = c("model", "vectors", "output"))

Arguments

model

an object inheriting from 'fasttext'.

file

a character string giving the name of the file.

what

a character string giving what should be saved.

Examples

## Not run: 
ft_save(model, "my_model.bin", what = "model")

## End(Not run)

Get Sentence Vectors

Description

Obtain sentence vectors from a previously trained model.

Usage

ft_sentence_vectors(model, sentences)

Arguments

model

an object inheriting from "fasttext".

sentences

a character vector giving the sentences.

Value

a matrix containing the sentence vectors.

Examples

## Not run: 
ft_sentence_vectors(model, c("sentence", "vector"))

## End(Not run)

Evaluate the Model

Description

Evaluate the quality of the predictions. For the model evaluation precision and recall are used.

Usage

ft_test(model, file, k = 1L, threshold = 0)

Arguments

model

an object inheriting from 'fasttext'.

file

a character string giving the location of the validation file.

k

an integer giving the number of labels to be returned.

threshold

a double giving the threshold.

Examples

## Not run: 
ft_test(model, file)

## End(Not run)

Train a Model

Description

Train a new word representation model or supervised classification model.

Usage

ft_train(
  file,
  method = c("supervised", "cbow", "skipgram"),
  control = ft_control(),
  ...
)

Arguments

file

a character string giving the location of the input file.

method

a character string giving the method, possible values are 'supervised', 'cbow' and 'skipgram'.

control

a list giving the control variables, for more information see ft_control.

...

additional control arguments inserted into the control list.

Examples

## Not run: 
cntrl <- ft_control(nthreads = 1L)
model <- ft_train("my_data.txt", method="supervised", control = cntrl)

## End(Not run)

Get Word Vectors

Description

Obtain word vectors from a previously trained model.

Usage

ft_word_vectors(model, words)

Arguments

model

an object inheriting from "fasttext".

words

a character vector giving the words.

Value

a matrix containing the word vectors.

Examples

## Not run: 
ft_word_vectors(model, c("word", "vector"))

## End(Not run)

Get Words

Description

Obtain all the words from a previously trained model.

Usage

ft_words(model)

Arguments

model

an object inheriting from "fasttext".

Value

a character vector.

Examples

## Not run: 
ft_words(model)

## End(Not run)

Predict using a Previously Trained Model

Description

Predict values based on a previously trained model.

Usage

ft_predict(
  model,
  newdata,
  k = 1L,
  threshold = 0,
  rval = c("sparse", "dense", "slam"),
  ...
)

Arguments

model

an object inheriting from 'fasttext'.

newdata

a character vector giving the new data.

k

an integer giving the number of labels to be returned.

threshold

a double withing [0, 1] giving lower bound on the probabilities. Predictions with probabilities below this lower bound are not returned. The default is 0 which means all predictions are returned.

rval

a character string controlling the return value, allowed values are "sparse", "dense" and "slam". The default is sparse, here the values are returned as a data.frame in a format similar to a simple triplet matrix (sometimes refereed to as the coordinate format). If rval is set to "dense", a matrix of the probabilities is returned. Similarly if rval is set to "slam", a matrix in the simple triplet sparse format from the slam package is returned.

...

currently not used.

Value

NULL if a 'result_file' is given otherwise if 'prob' is true a data.frame with the predicted labels and the corresponding probabilities, if 'prob' is false a character vector with the predicted labels.

Examples

## Not run: 
ft_predict(model, newdata)

## End(Not run)