Базовая линия тегирования сущностей

AbstractWe propose a two-stage neural model to tackle question generation from documents. First, our model estimates the probability that word sequences in a document are ones that a human would …...