Scidata.IMDBReviews (Scidata v0.1.5) View Source

Module for downloading the Large Movie Review Dataset.

Link to this section Summary

Functions

Downloads the IMDB reviews training dataset or fetches it locally.

Downloads the IMDB reviews test dataset or fetches it locally.

Link to this section Types

Specs

opts() :: [
  transform_inputs: ([binary(), ...] -> any()),
  transform_labels: ([integer(), ...] -> any())
]

Specs

test_sentiment() :: :pos | :neg

Specs

train_sentiment() :: :pos | :neg | :unsup

Link to this section Functions

Specs

download([{:example_types, [test_sentiment()]}]) :: %{
  review: [binary(), ...],
  sentiment: 1 | 0
}

Downloads the IMDB reviews training dataset or fetches it locally.

example_types specifies which examples in the dataset should be returned according to each example's label: :pos for positive examples, :neg for negative examples, and :unsup for unlabeled examples.

Link to this function

download_test(opts \\ [])

View Source

Specs

download_test([{:example_types, [test_sentiment()]}]) :: %{
  review: [binary(), ...],
  sentiment: 1 | 0
}

Downloads the IMDB reviews test dataset or fetches it locally.

example_types is the same argument in download/2 but excludes :unsup because all unlabeled examples are in the training set.