Deep learning of the natural language processing

During the last decades, artificial neural networks and deep learning approaches have strongly renewed the research perspectives in Natural Language Processing (NLP). Neural networks provide an efficient way for representation learning, yielding important improvement in several tasks. These tasks include document classification, syntactic parsing, automatic speech recognition and machine translation. Whereas the performances achieved by neural networks are impressive, their conception and optimization are still challenging. Moreover, these architectures are merely understood as efficient black boxes and their results remain difficult to interpret and explain. This special issue explores contributions of deep learning to NLP, their promises along with their limits and peculiarities.