1 paper
Jón Friðrik Daðason, David Erik Mollberg, Hrafn Loftsson +1
In this paper, we present a character-based BiLSTM model for splitting Icelandic compound words, and show how varying amounts of training data affects the performance of the model.…