Text data augmentations: Permutation, antonyms and negation

Text has traditionally been used to train automated classifiers for a multitude of purposes, such as: classification, topic modelling and sentiment analysis. State-of-the-art LSTM classifier require large number training examples avoid biases successfully generalise. Labelled data greatly improves classification results, but not all modern datasets include numbers labelled examples. Labelling is complex task that can be expensive, time-consuming, potentially introduces biases. Data augmentation methods create synthetic based on existing examples, with the goal improving results. These have in image tasks recent research extended them text classification. We propose method uses sentence permutations augment an initial dataset, while retaining key statistical properties dataset. evaluate our eight different baseline Deep Learning process. This permutation significantly accuracy by average 4.1%. also two more augmentations reverse each augmented example, antonym negation. test these three eligible datasets, results suggest -averaged, across datasets-improvement 0.35% 0.4% negation, when compared proposed augmentation.