language models are at the heart of numerous works, notably in the text
mining and information retrieval communities. These statistical models aim at
extracting word distributions, from simple unigram models to recurrent
approaches with latent variables that capture subtle dependencies