Набор данных 5

A large number of Open Relation Extrac- tion approaches have been proposed recently, covering a wide range of NLP machinery, from “shallow” (e.g., part-of-speech tagging) to “deep” (e.g., semantic role …...

AbstractWe study a search-based paraphrase generation scheme where candidate paraphrases are generated by iterated transformations from the original sentence and evaluated in terms of syntax quality, semantic distance, and lexical …...

AbstractTraining mixed-domain translation models is a complex task that demands tailored architec- tures and costly data preparation techniques. In this work, we leverage federated learning (FL) in order to tackle …...