Robust-to-Noise Models in Natural Language Processing Tasks

There are a lot of noisy texts surrounding a person in modern life. A traditional approach is to use spelling correction, yet the existing solutions are far from perfect. We propose a robust to noise word embeddings model which outperforms existing commonly used models like fasttext and word2vec in different tasks. In addition, we investigate the noise robustness of current models in different natural language processing tasks. We propose extensions for modern models in three downstream tasks, i.e. text classification, named entity recognition and aspect extraction, these extensions show improvement in noise robustness over existing solutions.