Skip to contents

Get Text Embeddings by Averaging Word Embeddings

embed_docs.Rd

textstat_embedding() takes a 'quanteda' dfm. embed_docs() is a more versatile function for which acts directly on either a character vector or a column of texts in a dataframe.

Usage

embed_docs(x, ...)

# Default S3 method
embed_docs(
  x,
  model,
  w = NULL,
  method = "mean",
  ...,
  tolower = TRUE,
  output_embeddings = FALSE
)

# S3 method for class 'data.frame'
embed_docs(
  x,
  text_col,
  model,
  id_col = NULL,
  w = NULL,
  method = "mean",
  ...,
  .keep_all = FALSE,
  tolower = TRUE,
  output_embeddings = FALSE
)

textstat_embedding(
  dfm,
  model,
  w = NULL,
  method = "mean",
  output_embeddings = FALSE
)

Arguments

x

a character vector, a data frame, or data frame extension (e.g. a tibble)

...

additional parameters to pass to quanteda::tokens() or to the user-specified modeling function

model

an embeddings object. For embed_docs(), model can alternatively be a function that takes a character vector and outputs a dataframe with a row for each element of the input.

w

optional weighting for embeddings in model if model is an embeddings object. See average_embedding().

method

method to use for averaging. See average_embedding(). Note that method = "median" does not use matrix operations and may therefore be slow for datasets with many documents.

tolower

logical. Convert all text to lowercase? If model is an embeddings object, this value is passed to quanteda::dfm().

output_embeddings

FALSE (the default) returns a tibble. TRUE returns an embeddings object. See 'Value' for details.

text_col

string. a column of texts for which to compute embeddings

id_col

optional string. column of unique document ids

.keep_all

logical. Keep all columns from input? Ignored if output_embeddings = TRUE.

dfm

a quanteda dfm

Value

If output_embeddings = FALSE, a tibble with columns doc_id, and dim_1, dim_2, etc. or similar. If .keep_all = TRUE, the new columns will appear after existing ones. If output_embeddings = TRUE, an embeddings object with document ids as rownames.

See also

embed_tokens()

Examples

texts <- c("this says one thing", "and this says another")
texts_embeddings <- embed_docs(texts, glove_twitter_25d)
texts_embeddings
#> # A tibble: 2 × 26
#>   doc_id  dim_1 dim_2  dim_3  dim_4   dim_5  dim_6 dim_7  dim_8   dim_9  dim_10
#>   <chr>   <dbl> <dbl>  <dbl>  <dbl>   <dbl>  <dbl> <dbl>  <dbl>   <dbl>   <dbl>
#> 1 text1   0.114 0.167 0.180  -0.144 -0.0492 -0.465  1.77 -0.161 -0.414  -0.0989
#> 2 text2  -0.289 0.297 0.0145 -0.108 -0.248  -0.495  1.58 -0.234 -0.0946 -0.177 
#> # ℹ 15 more variables: dim_11 <dbl>, dim_12 <dbl>, dim_13 <dbl>, dim_14 <dbl>,
#> #   dim_15 <dbl>, dim_16 <dbl>, dim_17 <dbl>, dim_18 <dbl>, dim_19 <dbl>,
#> #   dim_20 <dbl>, dim_21 <dbl>, dim_22 <dbl>, dim_23 <dbl>, dim_24 <dbl>,
#> #   dim_25 <dbl>

# quanteda workflow
library(quanteda)
#> Package version: 4.3.1
#> Unicode version: 15.1
#> ICU version: 74.2
#> Parallel computing: disabled
#> See https://quanteda.io for tutorials and examples.
texts_dfm <- dfm(tokens(texts))

texts_embeddings <- textstat_embedding(texts_dfm, glove_twitter_25d)
texts_embeddings
#> # A tibble: 2 × 26
#>   doc_id  dim_1 dim_2  dim_3  dim_4   dim_5  dim_6 dim_7  dim_8   dim_9  dim_10
#>   <chr>   <dbl> <dbl>  <dbl>  <dbl>   <dbl>  <dbl> <dbl>  <dbl>   <dbl>   <dbl>
#> 1 text1   0.114 0.167 0.180  -0.144 -0.0492 -0.465  1.77 -0.161 -0.414  -0.0989
#> 2 text2  -0.289 0.297 0.0145 -0.108 -0.248  -0.495  1.58 -0.234 -0.0946 -0.177 
#> # ℹ 15 more variables: dim_11 <dbl>, dim_12 <dbl>, dim_13 <dbl>, dim_14 <dbl>,
#> #   dim_15 <dbl>, dim_16 <dbl>, dim_17 <dbl>, dim_18 <dbl>, dim_19 <dbl>,
#> #   dim_20 <dbl>, dim_21 <dbl>, dim_22 <dbl>, dim_23 <dbl>, dim_24 <dbl>,
#> #   dim_25 <dbl>

# dplyr workflow
texts_df <- data.frame(text = texts)
texts_embeddings <- texts_df |> embed_docs("text", glove_twitter_25d)
texts_embeddings
#> # A tibble: 2 × 26
#>   doc_id  dim_1 dim_2  dim_3  dim_4   dim_5  dim_6 dim_7  dim_8   dim_9  dim_10
#>   <chr>   <dbl> <dbl>  <dbl>  <dbl>   <dbl>  <dbl> <dbl>  <dbl>   <dbl>   <dbl>
#> 1 text1   0.114 0.167 0.180  -0.144 -0.0492 -0.465  1.77 -0.161 -0.414  -0.0989
#> 2 text2  -0.289 0.297 0.0145 -0.108 -0.248  -0.495  1.58 -0.234 -0.0946 -0.177 
#> # ℹ 15 more variables: dim_11 <dbl>, dim_12 <dbl>, dim_13 <dbl>, dim_14 <dbl>,
#> #   dim_15 <dbl>, dim_16 <dbl>, dim_17 <dbl>, dim_18 <dbl>, dim_19 <dbl>,
#> #   dim_20 <dbl>, dim_21 <dbl>, dim_22 <dbl>, dim_23 <dbl>, dim_24 <dbl>,
#> #   dim_25 <dbl>