nlpaug 0.0.7

Natural language processing augmentation library for deep neural networks

nlpaug

This python library helps you with augmenting nlp for your machine learning projects. Visit this introduction to understand about Data Augmentation in NLP. Augmenter is the basic element of augmentation while Flow is a pipeline to orchestra multi augmenter together.

• Data Augmentation library for Text
• Data Augmentation library for Speech Recognition
• Data Augmentation library for Audio
• Does your NLP model able to prevent adversarial attack?

Starter Guides

• Augmentation for character and word
• Augmentation for spectrogram (audio input)
• Augmentation for audio
• How to train TF-IDF model
• How to create custom augmentation
• API Documentation

Augmenter

Target	Augmenter	Action	Description
Character	RandomAug	insert	Insert character randomly
		substitute	Substitute character randomly
		swap	Swap character randomly
		delete	Delete character randomly
	OcrAug	substitute	Simulate OCR engine error
	KeyboardAug	substitute	Simulate keyboard distance error
Word	RandomWordAug	swap	Swap word randomly
		delete	Delete word randomly
	SpellingAug	substitute	Substitute word according to spelling mistake dictionary
	WordNetAug	substitute	Substitute word according to WordNet's synonym
	WordEmbsAug	insert	Insert word randomly from word2vec, GloVe or fasttext dictionary
		substitute	Substitute word based on word2vec, GloVe or fasttext embeddings
	TfIdfAug	insert	Insert word randomly trained TF-IDF model
		substitute	Substitute word based on TF-IDF score
	BertAug	insert	Insert word based by feeding surroundings word to BERT language model
		substitute	Substitute word based by feeding surroundings word to BERT language model
Spectrogram	FrequencyMaskingAug	substitute	Set block of values to zero according to frequency dimension
	TimeMaskingAug	substitute	Set block of values to zero according to time dimension
Audio	NoiseAug	substitute	Inject noise
	PitchAug	substitute	Adjust audio's pitch
	ShiftAug	substitute	Shift time dimension forward/ backward
	SpeedAug	substitute	Adjust audio's speed
	CropAug	delete	Delete audio's segment
	LoudnessAug	substitute	Adjust audio's volume
	MaskAug	substitute	Mask audio's segment

Flow

Pipeline	Description
Sequential	Apply list of augmentation functions sequentially
Sometimes	Apply some augmentation functions randomly

Installation

The library supports python 3.5+ in linux and window platform.

To install the library:

pip install nlpaug

or install the latest version (include BETA features) from github directly

pip install git+https://github.com/makcedward/nlpaug.git

If you use BertAug, install the following dependencies as well

pip install pytorch_pretrained_bert torch

If you use WordEmbsAug (word2vec, glove or fasttext), downloading pre-trained model first

from nlpaug.util.file.download import DownloadUtil
DownloadUtil.download_word2vec(dest_dir='.') # Download word2vec model
DownloadUtil.download_glove(model_name='glove.6B', dest_dir='.') # Download GloVe model
DownloadUtil.download_fasttext(model_name='wiki-news-300d-1M', dest_dir='.') # Download fasttext model

Recent Changes

BETA Aug 16, 2019

• Add new augmenter (CropAug, LoudnessAug, MaskAug)
• QwertyAug is deprecated. It will be replaced by KeyboardAug
• Remove StopWordsAug. It will be replaced by RandomWordAug
• Code refactoring
• Added model download function for word2vec, GloVe and fasttext

0.0.6 Jul 29, 2019:

• Added new augmenter TF-IDF based word replacement augmenter(TfIdfAug)
• Added new augmenter Spelling mistake simulation augmenter(SpellingAug)
• Added new augmenter Stopword Dropout augmenter(StopWordsAug)
• Fixed #14

See changelog for more details.

Test

Word2vec, GloVe, Fasttext models are used in word insertion and substitution. Those model files are necessary in order to run test case. You have to add ".env" file in root directory and the content should be
    -   MODEL_DIR={MODEL FILE PATH}

Folder structure of model should be
    -- root directory
        - glove.6B.50d.txt
        - GoogleNews-vectors-negative300.bin
        - wiki-news-300d-1M.vec

Research Reference

Some of the above augmenters are inspired by the following research papers. However, it does not always follow original implementation due to different reasons. If original implementation is needed, please refer to original source code.

Augmenter	Inspired by
RandomAug, SpellingAug	Y. Belinkov and Y. Bisk. Synthetic and Natural Noise Both Break Neural Machine Translation. 2017
RandomAug	J. Ebrahimi, A. Rao, D. Lowd and D. Dou. HotFlip: White-Box Adversarial Examples for Text Classification. 2018
RandomAug, RandomWordAug	J. Ebrahimi, D. Lowd and Dou. On Adversarial Examples for Character-Level Neural Machine Translation. 2018
RandomAug, KeyboardAug	D. Pruthi, B. Dhingra and Z. C. Lipton. Combating Adversarial Misspellings with Robust Word Recognition. 2019
RandomAug, RandomWordAug	T. Niu and M. Bansal. Adversarial Over-Sensitivity and Over-Stability Strategies for Dialogue Models. 2018
RandomWordAug, WordNetAug	P. Minervini and S. Riedel. Adversarially Regularising Neural NLI Models to Integrate Logical Background Knowledge. 2018
WordNetAug	X. Zhang, J. Zhao and Y. LeCun. Character-level Convolutional Networks for Text Classification. 2015
WordNetAug	S. Kobayashi and C. Coulombe. Text Data Augmentation Made Simple By Leveraging NLP Cloud APIs. 2018
TfIdfAug	Q. Xie, Z. Dai, E Hovy, M. T. Luong and Q. V. Le. Unsupervised Data Augmentation. 2019
WordEmbsAug	W. Y. Wang and D. Yang. That’s So Annoying!!!: A Lexical and Frame-Semantic Embedding Based Data Augmentation Approach to Automatic Categorization of Annoying Behaviors using #petpeeve Tweets. 2015
BertAug	S. Kobayashi. Contextual Augmentation: Data Augmentation by Words with Paradigmatic Relation. 2018
FrequencyMaskingAug, TimeMaskingAug	D. S. Park, W. Chan, Y. Zhang, C. C. Chiu, B. Zoph, E. D. Cubuk and Q. V. Le. SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition. 2019

Data Source

Capatured data from internet for building augmenter/ test case.

See data source for more details.