Overview of Augmenters#

The following tables list all the available augmenters in augmenty, along with a short description. It also contains list all of the labels which the augmentersrespects. For instance, if you wish to train a named entity recognition pipeline you should not use augmenters which do not respect entity labels. Similarly, a hint is also given to whether the augmenter is recommended for training or evaluation. Lastly, the package includes a list of references to any data or packages used as well as references to example application of the augmenter in practice.

Augmenter name

Description

Token

Dependency parsing

Entity

Document

Training

Evaluation

References

char_replace_random_v1

Creates an augmenter that replaces a character with a random character from the keyboard.

✅

✅

✅

✅

✅

✅

char_replace_v1

Creates an augmenter that replaces a character with a random character from replace dict.

✅

✅

✅

✅

✅

✅

char_swap_v1

Creates an augmenter that swaps two neighbouring characters in a token with a given probability.

✅

✅

✅

✅

✅

✅

conditional_token_casing_v1

Creates an augmenter that conditionally cases the first letter of a token based on the getter. Either lower og upper needs to specifiedd as True.

✅

✅

✅

✅

✅

✅

da_historical_noun_casing_v1

Creates an augmenter that capitalizes nouns.

✅

✅

✅

✅

✅

✅

da_æøå_replace_v1

Creates an augmenter that augments æ, ø, and å into their spelling variants ae, oe, aa.

✅

✅

✅

✅

✅

✅

duplicate_token_v1

Creates an augmenter that randomly duplicate a token token.

✅

❌

✅

✅

✅

✅

ents_format_v1

Creates an augmenter which reorders and formats a entity according to reordering and formatting functions.

✅

✅

✅

✅

✅

✅

ents_replace_v1

Create an augmenter which replaces an entity based on a dictionary lookup.

✅

✅

✅

✅

✅

✅

keystroke_error_v1

Creates a augmenter which augments a text with plausible typos based on keyboard distance.

✅

✅

✅

✅

✅

✅

letter_spacing_augmenter_v1

Typically casing is used to add emphasis to words, but letter spacing has also been used to add e m p h a s i s to words (e.g. by Grundtvig; Baunvig, Jarvis and Nielbo, 2020). This augmenter randomly adds letter spacing emphasis to words. This augmentation which are human readable, but which are clearly challenging for systems using a white-space centric tokenization.

✅

✅

✅

✅

✅

paragraph_subset_augmenter_v1

Create an augmenter that extracts a subset of a document.

✅

✅

✅

✅

✅

✅

per_replace_v1

Create an augmenter which replaces a name (PER) with a news sampled from the names dictionary.

✅

✅

✅

✅

✅

✅

random_casing_v1

Create an augment that randomly changes the casing of the document.

✅

✅

✅

✅

✅

✅

random_starting_case_v1

Creates an augmenter which randomly cases the first letter in each token.

✅

✅

✅

✅

✅

✅

random_synonym_insertion_v1

Creates an augmenter that randomly inserts a synonym or from the tokens context. The synonyms are based on wordnet.

✅

❌

✅

✅

✅

✅

remove_spacing_v1

Creates an augmenter that removes spacing with a given probability.

✅

✅

✅

✅

✅

spacing_insertion_v1

Creates and augmneter that randomly adds a space after a chara cter. Tokens are kept the same.

✅

✅

✅

✅

✅

spacy.combined_augmenter.v1

Create a data augmentation callback that uses orth-variant replacement. The callback can be added to a corpus or other data iterator during training.

✅

✅

✅

✅

✅

✅

spacy.lower_case.v1

Create a data augmentation callback that converts documents to lowercase. The callback can be added to a corpus or other data iterator during training.

✅

✅

✅

✅

✅

✅

spacy.orth_variants.v1

Create a data augmentation callback that uses orth-variant replacement. The callback can be added to a corpus or other data iterator during training.

✅

✅

✅

✅

✅

spongebob_v1

Create an augmneter that converts documents to SpOnGeBoB casing.

✅

✅

✅

✅

✅

token_dict_replace_v1

Creates an augmenter swaps a token with its synonym based on a dictionary.

✅

✅

✅

✅

✅

✅

token_insert_random_v1

Creates an augmenter that randomly swaps two neighbouring tokens.

✅

❌

✅

✅

✅

✅

token_insert_v1

Creates an augmenter that randomly inserts a token generated based on a insert function.

✅

❌

✅

✅

✅

✅

token_replace_v1

Creates an augmenter which replaces a token based on a replace function.

✅

✅

✅

✅

✅

✅

token_swap_v1

Creates an augmenter that randomly swaps two neighbouring tokens.

✅

✅

✅

✅

✅

✅

Usage: Wei and Zau (2019)

upper_case_v1

Create an augmenter that converts documents to uppercase.

✅

✅

✅

✅

✅

word_embedding_v1

Creates an augmenter which replaces a token based on a replace function.

✅

✅

✅

✅

✅

wordnet_synonym_v1

Creates an augmenter swaps a token with its synonym based on a dictionary.

✅

✅

✅

✅

✅

✅

Data: Miller (1998), Package: Steven (2006), Usage: Wei and Zau (2019)