Word segmentation plays a pivotal role in improving any Arabic NLPapplication. Therefore, a lot of research has been spent in improving itsaccuracy. Off-the-shelf tools, however, are: i) complicated to use and ii)domain/dialect dependent. We explore three language-independent alternatives tomorphological segmentation using: i) data-driven sub-word units, ii) charactersas a unit of learning, and iii) word embeddings learned using a character CNN(Convolution Neural Network). On the tasks of Machine Translation and POStagging, we found these methods to achieve close to, and occasionally surpassstate-of-the-art performance. In our analysis, we show that a neural machinetranslation system is sensitive to the ratio of source and target tokens, and aratio close to 1 or greater, gives optimal performance.
展开▼