Reordering Symbols: A Step Towards a Symbol to Arabic Text Translator

Graphic symbols can be used as an alternative way of communication. Translating from a message composed of symbols to a fluent sentence will enable symbol users to be understood by those who may not be familiar with the use of symbols. Symbol messages may not match the target language in terms of order and syntax. This paper describes an attempt to reorder symbols, based on their labels, to match the target language namely Modern Standard Arabic. An initial experiment has been conducted using a SMT decoder with two n-gram models to reorder words in general then discuss its application to symbols. The output was evaluated using BLEU: an automatic evaluation metric used in machine translation. The average score of the output has improved over the input. Further improvements are suggested and will be carried out in future experiments.